Saturday, February 28, 2009

Looking for "Bad Stuff", pt II

After part I of what looks like it may become a series of Looking for "Bad Stuff" posts, I thought it would be a good idea to address this topic a bit more; clearly, one of the biggest issues most analysts may have, regardless of affiliation (LE, corporate consultant, etc.), will be simply where to begin analysis in the absence of specific guidance or criteria. Sometimes even repeated and detailed interviews of IT staff do not provide you with the information you need (or worse, may send you off in the wrong direction) and hence, you need to start by casting a wide net through malware scans (AV, anti-spyware, rootkit detectors, etc.).

So, in an attempt to develop a codified process as a response to this question, we need to start by addressing a couple of things. First, at some point before you actually start performing analysis, even before you begin the engagement, you need to ask yourself, What do I hope to achieve? Many times, this may be defined, to some degree, for you by a customer...other times, it may not. Once you start to understand your own goals, you then need to ask, What data do I need to achieve it? Whether you're beginning an engagement and scoping your acquisition, or you're sitting down to begin analysis of a single acquired image, DOCUMENTING these questions and their answers is paramount! I know, I know..."you s*ck because you make me write stuff down!" Believe me, I've heard it all before...but the fact of the matter is, if you don't write it down, it didn't happen!

Casting Your Net
Once you're ready to kick off your analysis, there are a number of ways to get started. In my previous post, I mentioned mounting the acquired image as a read-only file system and scanning it with AV software. Chris added booting the acquired image with LiveView and scanning with live rootkit detection tools such as GMER. All of this is a great way to really cast a net, but the whole idea behind a net is that it's designed and created based on a current understanding of what you're trying to catch. Early fishermen created nets based on the size of the fish they were trying to catch, and anyone who watches Man vs Wild or Survivorman has seen Bear and Les try to catch pretty small stuff (really, just about anything). As we've seen with many of the malware outbreaks lately (really, over the years), some malware isn't detected by AV products until someone else finds it and submits a sample...so if multiple malware scans come up empty, don't think that this definitely means that there is NO malware on the system.

So, there's malware (AV, anti-spyware, etc.) scans, and there's also other ways to scan for unusual or suspicious files. In the previous post, I mentioned a couple of tools...missidentify, by Jesse Kornblum, and sigcheck, from MS/SysInternals. Both of these tools can be used to attempt to identify suspicious files on a system, particularly where executable files tend to reside (system32 dir, in a Program Files subdirectory, etc.). These tools are by no means definitive...they still require someone looking at the results to determine what's legit and what's not. The reason for this is that malware authors and intruders put a lot of time and effort into remaining persistent on systems, and a tool written six months ago to detect a specific set of techniques will no longer be sufficient. Besides, to be completely honest, in a great many engagements I've been on, the easiest thing to do has been to hide in plain sight.

Another tool I mentioned was WFPCheck...you'll notice that this tool doesn't have a link. This is something I wrote a while ago to help detect the presence of malware that subverts or disables Windows File Protection (WFP), and subsequently modifies "protected" files. Now, WFP is clearly not meant as a security or protection mechanism, and there are ways (albeit a finite number of them) to subvert or disable WFP (for example, see here and here); however, detecting the artifacts of an infection is sometimes the only way we have available of determining if there was an infection.

Speaking of artifacts, another means of getting started in our analysis that we can do in parallel with the malware scans is to extract important files (Registry hive files, EVT files) from the image before mounting it, and then do some targeted analysis of those files. Tools like RegRipper and the Evt2Xls tools are extremely valuable for this kind of work, in that they are fast, efficient, and depending on the user (or perhaps more accurately, the user community), can be very, very targeted. I've written a number of plugins over the past several months that look specifically for artifacts left by specific families of malware.

What NOT To Do
One of the most often used means of "analysis" that I've seen with customers (and in user forums...forii...whateva!!) is, "I found a file and searched for the file name on Google...". Folks, this is NOT an analysis technique. Sure, it's a way to start, but it should not be all you do. There are plenty of sources out there that provide a basic understanding of things like the PE header format, both on the web and in books (hint, hint). So all I'm sayin' here is, don't let a Google search for the file name or a string you found in the file be the end of your analysis.

4 comments:

Anonymous said...

Harlen, where can one download the additional plugins that you have written for malware detection?
Regards.

H. Carvey said...

To be honest, I haven't been providing the additional plugins b/c there doesn't seem to be any interest in them.

I monitor the RegRipper.net forums pretty consistently, and I've seen no requests there, nor in my inbox...

Anonymous said...

I can't believe this. The first thing I think about is "Where can i het this plugin?" Is it possbile you can upload the NEW plugins to the RR site after you write about them. Let the readers know they are there now for the taking after they have read the article. Regards.

H. Carvey said...

RegRipper hasn't been shareware or donationware...all I've ever asked for is feedback and input, so it's sort of been "feedback-ware". In the absence of any feedback or requests for these plugins, I haven't posted them.