Analyzing Msoffice Malware With Officemalscanner

Analyzing MSOffice malware with OfficeMalScanner

Version:

1.0

Last Update:

30th July 2009

Author:

Frank Boldewin / www.reconstructer.org

Analyzing MSOffice malware with OfficeMalScanner www.reconstructer.org File: Analyzing MSOffice malware with OfficeMalScanner.pdf

Page 1 of 19 30/07/2009

Table of Contents

1

ABSTRACT ................................................................. 3

2

INTRODUCTION TO OFFICEMALSCANNER .................. 4

3

FEATURE OVERVIEW ................................................. 5

4

PRACTICAL USAGE .................................................... 9

5

MALHOST-SETUP ......................................................15

6

CONCLUSION ............................................................19

7

REFERENCES .............................................................19

Analyzing MSOffice malware with OfficeMalScanner www.reconstructer.org File: Analyzing MSOffice malware with OfficeMalScanner.pdf

Page 2 of 19 30/07/2009

1

Abstract

"If you know the enemy and know yourself, your victory will not stand in doubt; if you know Heaven and know Earth, you may make your victory complete." Sun Tzu – Art of War If we believe statistics and trends from industry giants like Symantec or McAfee, the cybercrime sector is one of the most growing markets today. We all know the classic bank robbery is history and even attacks against the “system”, meaning governments, telecommunication infrastructure or energy plants are easier than ever before, with powerful malicious tools and botnets in hand. Further, browser-based attacks serving trojans who steal everything that matters, be it banking data, passwords, credit card infos, trustworthy documents or mail conversions, are being the order of the day. Next to attacks with standard exploit packs and well-known trojans, there are more and more targeted attacks, also known as spear phishing. Infamous examples for this are the espionage attacks against German-, UK- and US-Governments in September 2007 [1] or the GhostNet case in March 2009 [2]. While the normal attacks against machines connected to the internet are often done with so-called drive-by attacks, spear phishing is accomplished by sending authentic mails to a favoured few people. The mails contain attachments pretending to be imported presentations or whitepapers. If victims open these attachments, it is likely that they get owned with a special prepared exploit for standard tools like MSOffice or Adobe Reader. If the target is important enough and it is to be expected, that the security measures are high on the victim’s site, than so-called 0days (unknown exploitable bugs) are very likely to be deployed. Once installed on the victim’s boxes these trojans start stealing as much information as possible. If the fresh installed malware uses a well formed and proprietary rootkit it is getting hard to detect actions of the Trojan code. So, next to 0-day preventing measures like hardened OSes and applications it is important to have good forensic tools by hand, that gives the targeted site a powerful weapon to analyze the malicious gift. In the past, some good tools were developed by several people to analyze PDF documents, but there are still no good tools available for the MSOffice document format. This paper introduces a new tool called OfficeMalScanner, which aims to be a forensic tool for the MSOffice document format. Analyzing MSOffice malware with OfficeMalScanner www.reconstructer.org File: Analyzing MSOffice malware with OfficeMalScanner.pdf

Page 3 of 19 30/07/2009

2

Introduction to OfficeMalScanner

Some months ago, I had the need to analyze a malicious PowerPoint document and started searching for good tools on the web. What I was searching for was a document dissector, a shellcode fingerprint scanner and a VB macro detacher. The free tools I found were: • Officecat [3] - A command line utility that can be used to process Microsoft Office Documents for the presence of potential exploit conditions in the file. Unfortunately, this tool just prints out some infos if the scanned file is malicious or not and what exploit has been used (CVE and MS numbers). • STG Docfile Viewer [4] – This tool is a very basic format parser and was not much of help for my problem to get a clue where to find the shellcode and maybe its embedded encrypted executables. The next step I have started was to find some documentation about the file format itself and how to parse its structure. This time I had much more luck, as Microsoft was kind enough to release some very detailed papers about the format specs here [5] and here [6]. This helped me a lot to write my own forensic tool OfficeMalScanner [7]. In the following pages, I will describe in detail what can be done with this forensic utility. Be aware that the OfficeMalScanner only scans the older office binary file formats. Office 2007 and newer uses a XML based structure and it is very easy to look inside these files. You can open files with extensions like docx, pptx, xlsx, docm, pptm and so forth with WinZip or Winrar and then open these files again with a normal text editor to see what’s inside. Solitary exceptions are files containing VB macros (docm, pptm and xlsm). Next to the usual XML files, you should find a file called vbaproject.bin This file contains the compressed VB macrocode, which is not XML, but in old binary files format style. You can extract this file and then scan it with OfficeMalScanner to uncompress the VB macrocode data. However, this will be described in detail later in this paper. The last notable tool before we start comes from Microsoft and was released only some days ago. It is called OffVis [8] and is a very nice MSOffice file format defragmentation util. Even if it is still in a “Beta” status, yet I suggest you to give it a try as well.

Analyzing MSOffice malware with OfficeMalScanner www.reconstructer.org File: Analyzing MSOffice malware with OfficeMalScanner.pdf

Page 4 of 19 30/07/2009

3

Feature Overview

Before I show you the practical usage of OfficeMalScanner check out the features below: The “SCAN” feature scans the entire malicious file for generic shellcode patterns. Here is a list of all currently implemented checks. GetEIP (4 Methods) CALL NEXT NEXT: POP reg ------------------------------------------JMP [0xEB] 1ST 2ND: POP reg 1ST: CALL 2ND ------------------------------------------JMP [0xE9] 1ST 2ND: POP reg 1ST: CALL 2ND ------------------------------------------FLDZ FSTENV [esp-0ch] POP reg Find Kernel32 base (3 methods) MOV reg, DWORD PTR FS:[30h] --------------------------------------------XOR reg_a,reg_a MOV reg_a(low-byte), 30h MOV reg_b, fs:[reg_a] --------------------------------------------PUSH 30h POP reg_a MOV reg_b, FS:[reg_a]

Analyzing MSOffice malware with OfficeMalScanner www.reconstructer.org File: Analyzing MSOffice malware with OfficeMalScanner.pdf

Page 5 of 19 30/07/2009

API Hashing LOOP: LODSB TEST al, al JZ short OK ROR EDI, 0Dh or ROR EDI, 07h ADD EDI, EAX JMP short LOOP OK: CMP EDI, ... Indirect function call PUSH DWORD PTR [EBP+val] CALL[EBP+val] Suspicious strings UrlDownloadToFile GetTempPath GetWindowsDirectory GetSystemDirectory WinExec IsBadReadPtr IsBadWritePtr CreateFile CloseHandle ReadFile WriteFile SetFilePointer VirtualAlloc GetProcAddr LoadLibrary

Analyzing MSOffice malware with OfficeMalScanner www.reconstructer.org File: Analyzing MSOffice malware with OfficeMalScanner.pdf

Page 6 of 19 30/07/2009

Easy decryption trick LODS(x) XOR or ADD or SUB or ROL or ROR STOS(x) Embedded OLE Data (unencrypted) Signature: \xD0\xCF\x11\xE0\xA1\xB1\x1a\xE1 Function Prolog PUSH EBP MOV EBP, ESP SUB ESP, or ADD ESP, PE-File Signature (unencrypted) Offset 0x0 == MZ Offset 0x3c == e_lfanew Offset e_lfanew == PE The “BRUTE” feature is an easy XOR + ADD 0x00 – 0xFF buffer decryption. Every time a buffer is decrypted, the scanner looks for an embedded OLE signature or a valid PE-file. If it matches, both embedded OLE- and PE-files are saved to disk. The “DEBUG” mode displays found signatures as disassembly for detected code patterns and in hexview style for detected strings, OLE- and PE-files.

Analyzing MSOffice malware with OfficeMalScanner www.reconstructer.org File: Analyzing MSOffice malware with OfficeMalScanner.pdf

Page 7 of 19 30/07/2009

The malicious index rating can be used for automated analysis as threshold values. Every suspicious trace increases the malicious index counter depending on its hazard potential. INDEX Executables Code Strings OLE

SCORING 4 3 2 1

The “INFO” mode dumps OLE structures, offsets and length and saves found VB-Macro code to disk. To dump the VB macrocode to disk I use the Microsoft OLE API and some tricky parsing, as well as macro decompression by using the undocumented RtlDecompressBuffer() function from NTDLL.DLL If you want to scan an entire directory of malicious documents, use the small python script “SCANDIR.PY” supplied with the OfficeMalScanner package.

Analyzing MSOffice malware with OfficeMalScanner www.reconstructer.org File: Analyzing MSOffice malware with OfficeMalScanner.pdf

Page 8 of 19 30/07/2009

4

Practical usage

After a short overview of all features of the scanner, we are now ready to show its practical usage. The figure 4.1 shows the “USAGE” screen if you execute OfficeMalScanner without any parameters. Figure 4.1:

As you can see, the usage of this tool is very easy and next to the description of “options” and “switches”, also examples are given as well.

Analyzing MSOffice malware with OfficeMalScanner www.reconstructer.org File: Analyzing MSOffice malware with OfficeMalScanner.pdf

Page 9 of 19 30/07/2009

In Figure 4.2, we just used the “SCAN” option. The output tells us, that three different shellcode pattern types were found in the PowerPoint binary. The findings are all in a range between 0x300 bytes, which gives us a good feeling that this document is malicious. Figure 4.2:

Analyzing MSOffice malware with OfficeMalScanner www.reconstructer.org File: Analyzing MSOffice malware with OfficeMalScanner.pdf

Page 10 of 19 30/07/2009

To ensure that the scanner hasn’t only reported false positives we can use the “DEBUG” switch. In figure 4.3, we can clearly see that the scanner was right in its findings. Figure 4.3:

Analyzing MSOffice malware with OfficeMalScanner www.reconstructer.org File: Analyzing MSOffice malware with OfficeMalScanner.pdf

Page 11 of 19 30/07/2009

Now after we know that shellcode is included in this binary, we can check for more stuff, like encrypted embedded OLE data or PE-files. This is done by using the “BRUTE” option as seen in figure 4.4 Figure 4.4:

Analyzing MSOffice malware with OfficeMalScanner www.reconstructer.org File: Analyzing MSOffice malware with OfficeMalScanner.pdf

Page 12 of 19 30/07/2009

As we can see, the scanner found one embedded OLE file as well as three different PE-files, which were encrypted with the key XOR 0x85. After detection of these files, they were dumped to disk. The dumped embedded OLE file can be re-scanned now with OfficeMalScanner to find further malicious traces. The dumped PE-files can be loaded into IDA Pro or a debugger for a detailed analysis or uploaded to online analysis platforms like www.cwsandbox.org or www.virustotal.com now. Next to typical shellcode based MSOffice exploits there also exist malicious documents containing evil VB macrocode. To reveal such stuff we can use the “INFO” option as seen in figure 4.5 Figure 4.5:

The VB macrocode is then dumped into a separate directory where we can observe it with a normal text editor or a normal “type” command as shown in figure 4.6

Analyzing MSOffice malware with OfficeMalScanner www.reconstructer.org File: Analyzing MSOffice malware with OfficeMalScanner.pdf

Page 13 of 19 30/07/2009

Figure 4.6:

A short look reveals such an interesting “Private Sub Shellcode()” function, which stands for itself. It is fairly telltale that this VB macrocode is malicious. Furthermore the start bytes 77 90 are 0x4D 0x5A in hex and “MZ” in ASCII. This is another hint for a PE-File that might be dropped and executed by this macrocode. Watch the flash video “VB macrocode debugging” to learn how to debug such malicious codes and drop the PE-file in a save way.

Analyzing MSOffice malware with OfficeMalScanner www.reconstructer.org File: Analyzing MSOffice malware with OfficeMalScanner.pdf

Page 14 of 19 30/07/2009

5

MalHost-Setup

Another tool I have added to the OfficeMalScanner suite is called MalHostSetup. It is useful if you want to run the shellcode inside a malicious file from a specific offset. Especially for MSOffice exploits, I have seen shellcodes that work like the ones shown in Figure 5.1 Figure 5.1:

As you can see the shellcode enumerates through the file handle values, first tries to detect a valid file handle using the GetFileSize function, and if true, it checks for its file size. If it matches 0xec600, the shellcode knows he is inside itself and drops an encrypted PE-file from some offset and executes it. Usually some MSOffice executable like winword.exe or powerpoint.exe is the host for such documents, but if we try to avoid calling some MSOffice product for testing, we can use MalHost-Setup.

Analyzing MSOffice malware with OfficeMalScanner www.reconstructer.org File: Analyzing MSOffice malware with OfficeMalScanner.pdf

Page 15 of 19 30/07/2009

If we execute the utility without any parameters, we get a usage screen as shown in figure 5.2 Figure 5.2:

MalHost-Setup wants the malicious file as input-file, a name for the output-file that gets created and an offset where the shellcode starts. Furthermore, if we want to debug the shellcode, a “WAIT” feature was included which patches the beginning of the shellcode with 0xeb 0xfe (loop forever). However, this is optional.

Analyzing MSOffice malware with OfficeMalScanner www.reconstructer.org File: Analyzing MSOffice malware with OfficeMalScanner.pdf

Page 16 of 19 30/07/2009

To get a clue were the shellcode starts we often have a good orientation point when we use OfficeMalScanner before and shellcode patterns like FS:30 (find kernel-base trick) were found, because if we look from the found offset some instructions above, the chance is very high to find the start of the shellcode. To disassemble the malicious file very easily OfficeMalScanner supplies the analyst with a small tool called DisView. See Figure 5.3 how use it. Figure 5.3:

Analyzing MSOffice malware with OfficeMalScanner www.reconstructer.org File: Analyzing MSOffice malware with OfficeMalScanner.pdf

Page 17 of 19 30/07/2009

If we found the right offset to the shellcode-start, we are ready to fire off MalHost-Setup as shown in figure 5.4 Figure 5.4:

MalHost-Setup then takes the malicious file, attaches it as overlay behind its execution engine and stores it on disk as separate output file. If you execute outfile.exe now, be aware to start it in a safe environment, otherwise it is likely that your system might be infected, if the shellcodestart you selected was the right one. ;-) As already mentioned above you can use the “WAIT” option to patch the shellcode-start. Be sure to write down the original bytes MalHost-Setup prints out to console for re-patching in the debugging session (see figure 5.5). If you start the 0xeb 0xfe patched outfile.exe now, it loops forever and waits for debugger attaching, e.g. in Ollydbg just attach to outfile.exe, press the “run” button and right after this the “pause” button and you should be at the 0xeb 0xfe loop. Now just re-patch the bytes to the original ones and start your debugging session. Figure 5.5:

Analyzing MSOffice malware with OfficeMalScanner www.reconstructer.org File: Analyzing MSOffice malware with OfficeMalScanner.pdf

Page 18 of 19 30/07/2009

6

Conclusion

With OfficeMalScanner, you got a tool to do forensics on MSOffice files, which might be malicious even if I tested the scanner successfully with thousands of malicious samples, it should be clear, that the bad guys still might use more heavy obfuscation tricks in future, to avoid generic shellcode detection. So if you find malicious samples on the web, which OfficeMalScanner detects as “clean”, do not hesitate to send them to me and I will try to find a generic way for detection. Next to this, future releases will contain more effective crypto-analysis tricks to detect encrypted PE-files. Further keep in mind, that this software was written in the C-language, hence my code also might contain exploitable bugs. So if you work with this tool, ensure you only use it in a safe environment. Suggestions and constructive reviews are always welcome.

7

References

[1] The GhostNet case

http://en.wikipedia.org/wiki/GhostNet

[2] Chinese hacked into Pentagon

http://www.ft.com/cms/s/0/9dba9ba2-5a3b-11dc-9bcd-0000779fd2ac.html

[3] Officecat

http://www.snort.org/vrt/vrt-resources/officecat

[4] STG Docfile Viewer

http://support.microsoft.com/?scid=kb%3Ben-us%3B139545&x=16&y=9

[5] Microsoft Office Binary (doc, xls, ppt) File Formats http://www.microsoft.com/interop/docs/OfficeBinaryFormats.mspx

[6] Microsoft Office Word 97-2007 binary file format specification http://download.microsoft.com/download/0/B/E/0BE8BDD7-E5E8-422A-ABFD4342ED7AD886/WindowsCompoundBinaryFileFormatSpecification.pdf

[7] OfficeMalScanner

www.reconstructer.org/code/OfficeMalScanner.zip

[8] OffVis

http://www.microsoft.com/downloads/details.aspx?displaylang=en&FamilyID=19a1a252c3af-4474-b33c-158c6e85115e

Thanks to Bruce Dang, Elia Florio, Michael Hale Ligh and Michael Sandee for suggestions and ideas.

Analyzing MSOffice malware with OfficeMalScanner www.reconstructer.org File: Analyzing MSOffice malware with OfficeMalScanner.pdf

Page 19 of 19 30/07/2009