Forensic Blogs

An aggregator for digital forensics blogs

by

Triaging with Tr3Secure Script’s NTFS Artifacts Only Option

An alert fires about an end point potentially having a security issue. The end point is not in the cubicle next to you, not down the hall, and not even in the same city. It's miles away in one of your organization's remote locations. Or maybe the end point is not miles away but a few floors from you. However, despite its closeness the system has a one terabyte hard drive in it. The alert fired and the end point needs to be triaged but what options do you have. Do you spend the time to physically track down the end point, remove the hard drive, image the drive, and then start your analysis. How much time and resources would be spent approaching triage in this manner? How many other alerts would be overlooked while the focus is on just one? How much time and resources would be wasted if the alert was a false positive? In this post I'll demonstrate how the Tr3secure collection script can be leveraged to triage this type of alert.

Efficiency and Speed Matters
Triaging requires a delicate balance between thoroughness and speed. Too much time spent on one alert means not enough time for the other alerts. Time is not your friend when you are either trying to physically track down a system miles away or imaging a hard drive prior to analysis. Why travel miles when milliseconds will do?  Why collect terabytes/gigabytes when megabytes will do? Why spend time doing deep dive analysis until you confirm you have to? The answer to all these questions is to walk the delicate balance between thoroughness and speed. To quickly collect only the data you need to confirm if the alert is a security event or false positive. The tools one selects for triage is directly related to their ability to be thorough but fast. One crop of tools to help you strike this balance are triage scripts.

Triage scripts automate parts of the triage process; this may include either the data collection or analysis. The Tr3secure collection script automates the data collection. In my previous post - Tr3Secure Collection Script Updated  - I highlighted the new features I added in the script to collect the NTFS Change Journal ($UsnJrnl) and new menu option for only collecting NTFS artifacts. Triaging the alert I described previously can easily be done by leveraging Tr3Secure collection script's new menu option.

To demonstrate this capability I configure a virtual machine in an extremely vulnerable state then visited a few malicious sites to provide me with an end point to respond to.

Responding to the System
The command prompt was accessed on the target system. A drive was mapped to a network share to assist with the data collection. The network share not only is where the collected data will be stored but it is also where the Tr3Secure collect script resides. The command below uses the Tr3Secure collection script NTFS artifacts only option and stores the collected data in the network share with the drive (Y:). Disclaimer: leveraging network shares does pose a risk but it allows for the tools to be executed without being copied to the target and for the data to be stored in a remote location.


The image below shows part of the output from the command above.


The following is the data that was collected using Tr3secure collection script's NTFS artifacts only option:

     - Recentfilecach.bcf
     - Prefetch files
     - Master File Table ($MFT)
     - NTFS Change Journal ($UsnJrnl)
     - NTFS Logfile ($Logfile)

Triaging
Collecting the data is the first activity in triaging. The next step is to actually examine the data to confirm if the alert is an actual security event or a false positive. The triage process will vary based on what the event is. A fast spreading worm varies from a Trojan. Antivirus alerts vary from malicious network traffic (i.e. an end point communicating with a malicious IP). The scenario demonstrated in this post is for an alert firing that requires triaging an end point. There are certain indicators to look for when triaging end points for malware as outlined below:

     - Programs executing from temporary or cache folders
     - Programs executing from user profiles (AppData, Roaming, Local, etc)
     - Programs executing from C:ProgramData or All Users profile
     - Programs executing from C:RECYCLER
     - Programs stored as Alternate Data Streams (i.e. C:WindowsSystem32:svchost.exe)
     - Programs with random and unusual file names
     - Windows programs located in wrong folders (i.e. C:Windowssvchost.exe)
     - Other activity on the system around suspicious files

In addition to looking for the above indicators, it is necessary  to ask yourself a few important questions.

     1. What was occurring on the system around the time the alert generated?
     2. Is there any indicators in the alert to use examining the data?
     3. For any suspicious or identified malicious code, did it execute on the system?

The triage steps to use is dependent on the collected data. In this case, the data collected with NTFS artifacts option results in only examining: program execution and filesystem activity. These two examination steps are very effective for triaging end points as I previously illustrated in my post Triaging Malware Incidents.

Examining Program Execution Artifacts
The first artifact to check is the Recentfilecach.bcf to determine if any stand-alone programs executed on the endpoint. The file can be examined using a hex editor or Harlan's rfc.exe tool included in his WFA 4/e book materials. The command below parses this artifact:

rfc.exe C:Data-22WIN-556NOJB2SI8-11.04.14-19.36preserved-filesAppCompatRecentFileCache.bcf

The image below is highlighting a suspicious program. The program stands out since it is a program that executed from the lab user profile's temporary internet files folder. Right off the bat this artifact provides some useful information. The coffee.exe program executed within the past 24 hours, the program came from the Internet using the Internet Explorer web browser, and the activity is associated with the lab user profile.


The second program execution artifact collected during the NTFS artifacts only option are prefetch files. The command below parses this artifact with Winprefetchview:

winprefetchview.exe /folder C:Data-22WIN-556NOJB2SI8-11.04.14-19.36preserved-filesPrefetch

The parsed prefetch files were first sorted by process path to find any suspicious programs. Then they were sorted by last run time to identify any suspicious programs that executed recently or around the time of the alert. Nothing really suspicious jumped out in the parsed prefetch files. However, there was a svchost.exe prefetch file and this was examined since it may contain interesting file handles related to the suspicious program listed in the Recentfilecach.bcf. (to see why review my Recentfilecach.bcf post and pay close attention to where I discuss the process svchost.exe) This could be something but it could be nothing. The svchost.exe process had a file handle to the ad[2].htm file located in the temporarily internet files folder as shown below:


Examining Filesystem Artifacts
The Tr3Secure collection script collects the NTFS $MFT, $Logfile, and $UsnJrnl; all of which can provide a wealth of information. These artifacts can be examined in two ways. First is to parse them and then look at the activity around the time the alert fired. Second, is to parse them and then look at the activity around the time of the identified suspicious files appeared on the system. In this post I'm using the latter method since the alert is hypothetical.

The $MFT was parsed with Joakim Schicht's MFT2CSV program. The output format is in the log2timeline format to make it into a timeline.


The MFT2CSV csv file was imported into Excel and a search was performed for the file "coffee.exe". It's important to look at the activity prior to this indicator and afterwards to obtain more context about the event. As illustrated below, there was very little activity - besides Internet activity - around the time the coffee.exe file was created. This may not confirm how the file got there but it does help rule out certain attack avenues such as drive-by downloads.


The image below shows what occurred on the system next. There is activity of Java executing and this may appear to be related to a Java exploit. However, this activity was due to Java update program executing.


The last portion of the $MFT timeline is shown below.  A file with the .tmp extension was created in the temp directory followed by the creation of a file named SonicMaster.exe in the System32 folder. SonicMaster.exe automatically becomes suspicious due to its creation time around coffee.exe (guilty by association.) If this is related then that means the malware had administrative rights, which is required to make modifications to the System32 folder. The task manager is executed and then there was no more obvious suspicious activity in the rest of the timeline. The remaining activity was the lab user account browsing the Internet.


At this point the alert has been properly triaged and confirmed. The event can be escalated according to the incident response procedures. However, addition context can be obtained by parsing the NTFS Change Journal ($UsnJrnl). Collecting this artifact was the most recent update to the Tr3Secure collection script. The $UsnJrnl was parsed with Joakim Schicht's UsnJrnl2CSV program.


The UsnJrnl2CSV csv file was imported into Excel and a search was performed for the file "coffee.exe". It's important to look at the activity prior to this indicator and afterwards to obtain more context about the event.  The $UsnJrnl shows the coffee.exe file being dropped onto the system from the Internet (notice the [1] included in the filename.)


The next portion as shown below, shows coffee.exe executing, which resulted in the modification to the Recentfilecache.bcf. Following this the tmp file is created along with the SonicMaster.exe file. The interesting activity is the data overwrites made to the SonicMaster.exe file after it's created.


Immediately after the SonicMaster.exe file creation the data overwrites continued to other programs as illustrated below. To determine what is happening would require those executables to be examined but it appears they had data written to them.


The executables were initially missed in the $MFT timeline since the files by themselves were not suspicious. They were legitimate programs located in their correct folders. However, the parsed $UsnJrnl provided more context about the executables and circling back around to the $MFT timeline shows the activity involving them. Their $MFT entries were updated.


Conclusion
As I mentioned previously, triaging requires a delicate balance between thoroughness but speed. It's a delicate balance between quickly collecting and analyzing data from the systems involved with an alert to determine if the alert is a security event or false positive. This post highlighted an effective technique for triaging an end point and the tools one could use. The entire technique takes minutes to complete from beginning to end. The triage process did not confirm if the activity is a security event but it did determine additional time and resources needs to be spent digging a bit deeper. Doing so would had revealed that the coffee.exe file is malicious and is actually the Win32/Parite file infector. The malware infects all executables on the local drive and network shares. Digging even deeper by performing analysis on coffee.exe (Malwr report and Anubis report) matches the activity identified on the system and even provides more indicators (such as an IP address) one could use to search for other infections in the environment.

by

Timeline Analysis by Categories

Organizing is what you do before you do something, so that when you do it, it is not all mixed up.

~ A. A. Milne

"Corey, at times our auditors find fraud and when they do sometimes they need help collecting and analyzing the data on the computers and network. Could you look into this digital forensic thing just in case if something comes up?" This simple request - about seven years ago - is what lead me into the digital forensic and incident response field. One of the biggest challenges I faced was getting a handle on the process one should use. At the time there was a wealth of information about tools and artifacts but there was not as much outlining a process one could use. In hindsight, my approach to the problem was simplistic but very effective. I first documented the examination steps and then organized every artifact and piece of information I learned about beneath the appropriate step. Organizing my process in this manner enabled me to carry it out without getting lost in the data. In this post I'm highlighting how this type of organization is applied to timeline analysis leveraging Plaso.

Examinations Leveraging Categories
Organizing the digital forensic process by documenting the examination steps and organizing artifacts beneath them ensured I didn't get "all mixed up" when working cases. To do examination step "X" examine artifacts A, B, C, and D. Approaching cases this way is a more effective approach. Not only did it prevent jumping around but it helped to minimize overlooking or forgetting about artifacts of interest. Contrast this to a popular cheat sheet at the time I came into the field (links to cryptome). The cheat sheet was a great reference but the listed items are not in an order one can use to complete an examination. What ends up happening is that you jump around the document based on the examination step being completed. This is what tends to happen without any organization. Jumping around in various references depending on what step is being completed. Not an effective method.

Contrast this to the approach I took. Organizing every artifact and piece of information I learned about beneath the appropriate step. I have provided glimpses about this approach in my posts: Obtaining Information about the Operating System and It Is All About Program Execution. The "information about the operating system" post I wrote within the first year of starting this blog. In the post I outlined a few different examination steps and then listed some of the items beneath it. I did a similar post for program execution; beneath this step I listed various program execution artifacts. I was able to write the auto_rip program due to this organization where every registry artifact was organized beneath the appropriate examination step.

Examination Steps + Artifacts = Categories
What exactly are categories? I see them as the following: Examination Steps + Artifacts = Categories. I outlined my examination steps on the jIIr methodology webpage and below are some of the steps listed for system examinations.

        * Profile the System
              - General Operating System Information
              - User Account Information
              - Software Information
              - Networking Information
              - Storage Locations Information
        * Examine the Programs Ran on the System
        * Examine the Auto-start Locations
        * Examine Host Based Logs for Activity of Interest
        * Examine Web Browsing
        * Examine User Profiles of Interest
              - User Account Configuration Information
              - User Account General Activity
              - User Account Network Activity
              - User Account File/Folder Access Activity
              - User Account Virtualization Access Activity
        * Examine Communications

In a previous post, I said "taking a closer look at the above examination steps it’s easier to see how artifacts can be organized beneath them. Take for example the step Examine the programs ran on the system. Beneath this step you can organize different artifacts such as: application compatibility cache, userassist, and muicache. The same concept applies to every step and artifact." In essence, each examination step becomes a category containing artifacts. In the same post I continued by saying "when you start looking at all the artifacts within a category you get a more accurate picture and avoid overlooking artifacts when processing a case."

This level of organization is how categories can be leveraged during examinations.

Timeline Analysis Leveraging Categories
Organizing the digital forensic process by documenting the examination steps and organizing artifacts is not limited to completing the examination steps or registry analysis. The same approach works for timeline analysis. If I'm looking to build a timeline of a system I don't want everything (aka the kitchen sink approach.) I only want certain types of artifacts that I layer together into a timeline.

To illustrate let's use a system infected with commodity malware. The last thing I want to do is to create a supertimeline using the kitchen sink approach. First, it takes way too long to generate (I'd rather start analyzing than waiting.) Second, the end result has a ton of data that is really not needed. The better route is to select the categories of artifacts I want such as: file system metadata, program execution, browser history, and windows event logs. Approaching timelines in this manner makes them faster to create and easier to analyze.

The way I created timelines in this manner was not efficient as I wanted it to be. Basically, I looked at the timeline tools I used and what artifacts they supported. Then I organized the supported artifacts beneath the examination steps to make them into categories. When I created a timeline I would use different tools to parse the categorized artifacts I wanted and then combine the output into a single timeline in the same format. It sounds like a lot but it didn't take that long to create it. Hell, it was even a lot faster than doing the kitchen sink approach.

This is where Plaso comes into the picture by making things a lot easier to leverage categories in timeline analysis.

Plasm
Plaso is a collection of timeline analysis tools; one of which is plasm. Plasm is capable of "tagging events inside the plaso storage file." Translation: plasm organizes the artifacts into categories by tagging the event in the plaso storage file. At the time of this post the tagging occurs after the timeline data has already been parsed (as opposed to specifying the categories and only parsing those during timeline generation.) The plasm user guide webpage provides a wealth of information about the tool and how to use it. I won't rehash the basics since I couldn't do justice to what is already written. Instead I'll jump right in to how plasm makes leveraging categories in timeline analysis easier.

The plasm switch " --tagfile" enables a tag file to be used to define the events to tag. Plaso provides an example tag file named tag_windows.txt. This is the feature in Plaso that makes it easier to leverage categories in timeline analysis. The artifacts supported by Plaso are organized beneath the examination steps in the tag file. The image below is a portion of the tag_jiir.txt tag file I created showing the organization:


tag_jiir.txt is still a work in progress. As can be seen in the above image, the Plaso supported artifacts are organize beneath the " Examine the Programs Ran on the System" (program execution) and " Examine the Auto-start Locations" (autostarts info) examination steps. The rest of the tag file is not shown but the same organization was completed for the remaining examination steps. After the tag_jiir.txt tag file is applied to plaso storage file then timelines can be created only containing the artifacts within select categories.

Plasm in Action
It's easier to see plasm in action in order to fully explore how it helps with using categories in timeline analysis. The test system for this demonstration is one found laying around on my portable drive; it might be new material or something I previously blogged about. For demonstration purposes I'm running log2timeline against a forensic image instead of a mounted drive. Log2timeline works against a mounted drive but the timeline will lack creation dates (and at the time of this post there is not a way to bring in $MFT data into log2timeline.)

The first step is taking a quick look at the system for any indications that the system may be compromised.  When triaging for potential malware infected system the activity in the program execution artifacts excel and the prefetch files revealed a suspicious file as shown below.


The file stands out for a few reasons. It's a program executing from a temp folder inside a user profile. The file handle to the zone.identifier file indicates the file came from the Internet. Lastly, the program last ran on 1/16/14 12:32:14 UTC.

Now we move on to creating the timeline with the command below (the -o switch specifies the partition I want to parse.) FYI, the command below creates a kitchen sink timeline with everything being parsed. To avoid the kitchen sink use  plaso filters. Creating my filter is on my to-do list after I complete the tag_jiir.txt file.

log2timeline.exe -o 2048 C:Atadtest-datatest.dump C:Atadtest-datatest.E01

The image below shows the "information that have been collected and stored inside the storage container." The Plaso tool pinfo was used to show this information.


Now it is time to see plasm in action by tagging the events in the storage container. The command below shows how to do this using my tag_jiir.txt file.

plasm.exe tag --tagfile=C:Atadtest-datatag_jiir.txt C:Atadtest-datatest.dump

The storage container information now shows the tagged events as shown below. Notice how the events are actually the categories for my examination steps (keep in mind some categories are not present due to the image not containing the artifacts.)


Now a timeline can be exported from the storage container based on the categories I want. The program execution indicator revealed there may be malware and it came from the internet. The categories I would want for a quick look are the following:

        - Filesystem: to see what files were created and/or modified
        - Web browsing: to correlate web browsing activity with malware
        - Program execution: to see what programs executed (includes files such as prefetch as well as specific event logs

The command below creates my timeline based on the categories I want (-o switch outputs in l2tcsv format, -w switch outputs to a file, and filter uses the tag contains.) FYI, a timeslice was not used since I wanted to focus on the tagging capability.

psort.exe -o L2tcsv -w C:Atadtest-datatest-timeline.csv C:Atadtest-datatest.dump "tag contains 'Program Execution' or tag contains 'Filesystem Info' or tag contains 'Web Browsing Info'"

The image below shows the portion of the timeline around the time of interest, which was 1/16/14 12:32:14 UTC. The timeline shows the lab user account accessing their Yahoo web email then a file named invoice.83842.zip being downloaded from the internet. The user then proceeded to execute the archive's contents by clicking the invoice.83842.exe executable. This infection vector is clear as day since the timeline does not contain a lot of un-needed data.


Conclusion
Plaso's tagging capabilities makes things easier to leverage categories in timeline analysis. By producing a timeline containing only the categories one wants in order to view the timeline data for select artifacts. This type of organization helps minimize getting "all mixed up" when conducting timeline analysis by getting lost in the data.

Plaso is great and the tagging feature rocks but as I mentioned before I used a combination of tools to create timelines. Not every tool has the capabilities I need so combining them provides better results. In past I had excellent results leveraging the Perl log2timeline and Regripper to create timelines. At this point the tools are not compatible. Plaso doesn't have a TLN parser (can't read RegRipper's TLN plug-ins) and RegRipper only outputs to TLN. Based on Harlan's upcoming presentation my statement may not be valid for long.

In the end, leveraging categories in timeline analysis is very powerful. This train of thought is not unique to me. Others (who happen to be tool developers) are looking into this as well. Kristinn is as you can see in Plaso's support for tagging and Harlan wrote about this in his latest Windows forensic book.


Side note: the purpose of this post was to highlight Plaso's tagging capability. However, for the best results the filtering capability should be used to reduce what items get parsed in the first place. The kitchen sink approach just takes way too long; why wait when you can analyze.  

by

Malware Root Cause Analysis Dont Be a Bone Head Slide Deck

Today I gave a presentation titled Malware Root Cause Analysis Don't Be a Bone Head at the New York State Cyber Security Conference. This presentation was a follow-up to the presentation I gave last year titled Finding Malware Like Iron Man. Last year I laid out a triage process to find malware and this year I went into more depth discussing how the malware got there in the first place. This post contains the following for my talk: CFP, slide deck, and video I showed.

CFP
Computer users are confronted with a reoccurring issue every day. This happens regardless if the user is an employee doing work for their company or a person doing online shopping trying to catch the summer sales. The user is using their computer and the next thing you know it is infected with malware. Even Hollywood is not immune to this issue as illustrated in the TV show Bones.  The most common action to address a malware infection is to reimage, rebuild, and redeploy the system back into production.  Analysis of the system to understand where the malware came from is not a priority or goal.

Root case analysis needs to be performed on systems impacted by malware to improve decision making. The most crucial question to answer is how did this happen since it will determine if we were targeted and more importantly what can be done to mitigate this from re-occurring. Last year, in my presentation Finding Malware Like Iron Man I explored the first step in root cause analysis, which is locating the malware. The next step in root cause analysis is to identify the malware's source.

In this technical presentation Corey will discuss the root cause analysis process to determine how malware infected a computer running the Windows operating system. The topics will include: why perform root cause analysis, how not to perform root cause analysis,  compromise root cause analysis model, attack vector artifacts, and scenarios.

Slide Deck
Malware Root Cause Analysis Don't Be a Bone Head slides viewable online

Malware Root Cause Analysis Don't Be a Bone Head slides PDF file

Video
I chopped up the Bones TV episode The Crack in the Code (Season 7 episode 6) I purchased through iTunes. However, others have posted the segment I used in my presentation. For your viewing pleasure here is "Malware on Bone".