METHOD AND SYSTEM FOR AUTOMATIC DETECTION AND ANALYSIS OF MALWARE

[0001] This application claims the benefit of U.S. Provisional Application No. 61/377,427 filed Aug. 26, 2010, the disclosure of which is incorporated herein by reference.

[0002] The present disclosure relates generally to automatic detection and analysis of malware targeted to run on a computer system without informed consent.

[0003] Malware, short for malicious software, is software designed to infiltrate a computer system without the owner's informed consent. The expression is a general term used by computer professionals to mean a variety of forms of hostile, intrusive, or annoying software or program code. The term “computer virus” is sometimes used as a catch-all phrase to include all types of malware, including true viruses. Malware presents security issues for computer systems. Therefore, there is a need in the art for methods and systems for detecting and analyzing malware.

[0004] The present disclosure relates generally to computer networks. More specifically, the present disclosure relates to methods and systems for detecting and analyzing malware. The methods and techniques may be applied to a variety of computer networks and communications systems.

[0005] A method is provided to receive a file for analysis, store a memory baseline for a system, and copy the file to the system. The file is executed on the system, the operation of the system is terminated, and a post-execution memory map of the system is stored. The memory baseline and the post-execution memory map are analyzed. As a result of the analysis, the method may determine that the file is infected with malware.

[0006] Additional objects of the disclosure will be set forth in part in the description which follows, and in part will be obvious from the description, or may be learned by practice of embodiments consistent with the disclosure. It is to be understood that both the foregoing general description and the following detailed description are exemplary and explanatory only and are not restrictive of the application, as claimed.

[0007] The accompanying drawings, which are incorporated in and constitute a part of this specification, illustrate embodiments of the application and together with the description, serve to explain the principles of the application subject matter.

[0021] Embodiments of the present application provide automated malware analysis services to customers and internal research teams. The method, system, and computer-readable medium described herein rapidly produces reports that information customers may use to analyze threats and mitigate risks. As described more fully throughout the present specification, embodiments of the present application provide an automated system (with many other integrated components to be described herein) that uses forensic tools in a post-mortem manner to analyze the behavior of malicious code and/or detect malware. Such a system can be referred to as a sandbox, which, as will be evident to one of skill in the art, is a system in which malicious software can be executed in an environment designed to eliminate adverse effects on systems outside the sandbox. The product also performs a variety of dynamic and static analysis techniques to determine exactly what the malware does while running and its full range of capabilities.

[0022] Embodiments of the present application provide benefits not available using conventional techniques. As an example, the target environment (i.e., operating system on which the malicious code executes) is not altered in any way before the malware's execution. Conventional systems install kernel drivers, service programs, and alter system files in order to monitor how the malware behaves. In effect, these conventional products force malware to execute in a heavily modified, non-native environment, which can taint the results of the analysis. A particular example includes the Application Programming Interface (API) hooks used by conventional products to log the malware's interactions with the system. The API hooks actually act as “blinders” and prevent the conventional products from determining when malware installs its own API hooks for rootkit purposes. Embodiments of the present application execute malware on a target system and then automate a forensic investigation instead of using a pre-configured set of traps for the malware.

[0023] Additionally, the embodiments described herein include the capability to determine changes made to physical memory (RAM) as a result of executing malware. Embodiments utilize one or more advanced memory forensics platforms and ancillary software including databases to compute the difference between the baseline memory dump and the “infected” memory dump. By analyzing RAM, malicious behaviors are detected that conventional approaches cannot detect (such as the API hooks mentioned above), thus providing a more comprehensive and accurate report on the malware. Moreover, embodiments of the present application have the capability to perform in-depth analysis of rootkits running in kernel mode memory (referred to as the rootkit detection component). Conventional approaches, if they report on any kernel-level activity at all, are limited to a notification that a particular driver loaded. The forensics/analysis capability described herein allow the analysis to go much deeper into the heart of the operating system (while not being disruptive since operation occurs on a memory dump once the system is frozen or powered down) to detect attempts for malware to hide in the stealthiest ways.

[0024] Furthermore, embodiments track relationships between files and malware samples (referred to as the database component). A family tree-like structure is maintained, where the initial/original sample is the parent and any dropped or downloaded files are children. The children are analyzed individually and then all information is grouped together in a big-picture scenario. This allows customers to upload a group of files (potentially from the same security incident) and have them automatically linked together in the database. Additionally, embodiments of the present application are designed to operate with an interactive website, as opposed to just generating reports in HTML and PDF form that customers can view. As described below, users can “tag” elements of a report (such as a created file name, created registry key name, loaded DLL name, or the like) as suspicious and the website will highlight those elements on the report pages for other malware samples. Users can associate text-based notes or binary attachments with a report. For example, if a malware sample encrypted a file on a victim's hard drive, the encrypted file could be associated with the malware sample in the system's database. Later, when an analyst needs to investigate the malware, the analyst may access all required files from the same location—similar to a shared, network storage system.

[0025] FIG. 1A is an exemplary simplified schematic diagram of a malware detection and analysis system according to an embodiment. The high level architecture of embodiments of the present application includes a Database server, which can be a Linux system running MySQL. Other operating systems and other database systems are included within the scope of the present invention and Linux and MySQL are merely illustrated as an example. The malware detection and analysis controller provides for control of system operations. In the illustrated embodiment, the malware detection and analysis controller is based on a Linux system that stores the main code for automating each step of the analysis. It picks up any files in the queue and sends them to the virtual infrastructure or physical infrastructure (depending on the user's preferences).

[0026] As illustrated in FIG. 1A, the virtual-machine based infrastructure includes several virtual machine servers (e.g., VMware ESXi servers—each with 10-15 virtual machines running various versions of an operating systems (e.g., Windows XP, Windows Vista, Windows 7, or the like)). The ESXi servers store the virtual disk and memory files for the virtual machines on a SAN or NAS storage device, which is accessed by the malware detection and analysis controller during the analysis phase. To provide for network access, a web server is provided, which can be a Linux system running CGI scripts written in the Python programming language. The web server accesses the Database server over the network to insert malware samples provided in various manners as described below (e.g., samples that malware detection and analysis users have uploaded through the web form) and also to retrieve results of the analysis once complete. One of ordinary skill in the art that other virtual machine implementations may be used to achieve an architecture similar to the one described herein, as well as other known means of providing a programming interface on a webserver.

[0027] FIG. 1B is an exemplary simplified schematic diagram of a malware detection and analysis system according to another embodiment. As illustrated in FIG. 1B, the system receives one or more files and one or more preferences as inputs. Utilizing one or more components including a data processor, a pre-processor, virtual machine(s), physical machine(s), an emulator, a database, and/or an input/output module, the system detects malware and provides one or more reports as an output. The various components are not all required by embodiments of the present application and the system may use one or more of the illustrated components in performing the methods described herein. Further, additional components as described throughout the present specification can be utilized in place of or in conjunction with the illustrated components.

[0028] FIG. 2 is an exemplary simplified flowchart illustrating operation of the malware and analysis system according to an embodiment. FIG. 2 provides insight into the basic workflow of the system provided by embodiments of the present disclosure. Although multiple steps and processes are illustrated in FIG. 2, not all steps and processes are required by all embodiments. The description provided in relation to FIG. 2 is merely to exemplify various operations that are possible using embodiments of the present invention and it not intended to limit the scope of the present invention. Thus, it should be appreciated that the specific steps illustrated in FIG. 2 provide a particular method of detecting and analyzing malware according to an embodiment of the present disclosure. Other sequences of steps may also be performed according to alternative embodiments. For example, alternative embodiments may perform the steps outlined above in a different order. Moreover, the individual steps illustrated in FIG. 2 may include multiple sub-steps that may be performed in various sequences as appropriate to the individual step. Furthermore, additional steps may be added or removed depending on the particular applications. One of ordinary skill in the art would recognize many variations, modifications, and alternatives, including, without limitation, the ability to embody the system in instructions on a computer-readable medium, which when executed perform a method comprising a malware detection and analysis system.

[0029] Referring to FIG. 2, the system receives input(s) (e.g., malicious files) and a set of user preferences. The system performs an initial static analysis on the file or files and then dispatches the file or files to a physical machine, virtual machine, or emulator for dynamic analysis. The analysis engine uses forensic tools and techniques to determine what changed on the file system, registry, and in memory. The analysis engine scans for rootkits (described in more detail below in relation to the rootkit component) and inserts the data in the database. The website allows users to browse and interact with reports.

[0030] As described in additional detail in relation to FIG. 3 below, embodiments may provide detailed, accurate reports on malware, without modifying the operating system on which it runs, by extracting information (in some implementations, a large amount of information) about the state of the system (also referred to as the “baseline”) before executing the malware. The baseline includes the contents of the file system, registry, memory (RAM), and the like. After executing the malware, the system extracts a second information set (also referred to as the “comparison”) while the target operating system is powered down. The powering down of the target operating system prevents the malware from “fighting back” against the information gathering process. The system then computes the difference between the baseline and the comparison.

[0031] In order to implement the methods described herein, embodiments may utilize one or more of the following components: a web site that customers can log into to submit malware samples; a group of re-usable virtual machines and physical machines on which to execute the malware samples and/or other files; and a database for storage. In a particular implementation, a set of programs written in the Python programming language are executed on the malware detection and analysis controller illustrated in FIG. 1A.

[0032] FIG. 3 is an exemplary simplified flowchart illustrating a method of detecting and analyzing malware according to an embodiment. The method 300 includes receiving a file (310). The file, which may be one of several files is a potentially malicious file. The file may be any suitable file type that runs on an operating system (e.g., Microsoft Windows) and can include executables (.exe), dynamic link libraries (.dll), kernel drivers (.sys), Adobe Reader files (.pdf), Microsoft Office documents (.doc, .xls, .ppt), and URLs (typically, website addresses beginning with http://, https://, or ftp://). The file(s) may also include zip archives containing any number of the previously mentioned file types. The files can be received as a result of a customer submitting the files via one or more emails (which can include the files as attachments), uploading the one or more files to a website in communication with the system (e.g., through a web form), submitted by automated tools that are known as honeypots, which collect malware from various sites and then upload the files automatically to the system, or the like. One of ordinary skill in the art would recognize many variations, modifications, and alternatives. Thus, embodiments may utilize submissions by users, automatic submissions from honeypots or other suitable sources of malware and potential malware, and the like.

[0033] The file types include files that are executable on operating systems of interest. In an example, Windows-based operating systems are included within the scope of the present invention. In some embodiments, files that are suitable for operation on non-Windows-based operating systems are analyzed, including files that are executable on Linux-OS, Mac-OS, mobile operating systems, and the like. Thus, the embodiments have particular applicability to Windows-based operating systems, embodiments are not limited to these particular operating systems.

[0034] User preferences are also received by the system (312). Embodiments may provide a wide variety of parameters under which the submitted file will be run during the detection and analysis phases of operation. As an example, the customer or user can choose an operating system (e.g., Windows XP, Windows Vista, Windows 7, Windows Mobile, Windows Phone, or the like), a length of time to let the file execute, what privileges to use when executing the malware (administrator or non-administrator), whether the machine should be rebooted after installing the malware but before performing an analysis (some malware only activates after the first reboot), or whether the malware should have access to real or simulated data. In some embodiments, step 312 is optional and user preferences are obtained from data related to the user or some other suitable source.

[0035] FIG. 5 is an exemplary display screen of a graphical user interface according to an embodiment. In an embodiment, the user preferences are selected by using a web form similar to the form illustrated in FIG. 5. Referring to FIG. 5, a user is able to browse to and select a file for upload and specify a number of user preferences, which may have default values. As examples, a user may choose any argument to send to the file; an email address to which the user will receive a pdf copy of the analysis when the analysis is complete; a program to be run after the file is executed, which provides an opportunity to use additional types of analysis tools as desired by the user; the guest operating system (e.g., the malware can be run in Windows XP with Service Pack 2, Windows 7, or the like); whether the machine should reboot as sometimes malware does not activate until the first time the machine is rebooted; to run the malware as an administrative user; to run the malware as a non-administrative user (providing for an analysis of the same malware twice, once with administrative privileges and once with non-administrative privileges, thereby providing insight into what the malware does when it has administrator permissions as opposed to when it does not have these permissions); and other user preferences or parameters.

[0036] The runtime for the malware can be a useful user preference since, for some malware, the malware will not do anything malicious until a predetermined period of time after it is run for the first time, for example, five minutes or twenty minutes. In this case, when the user submits a file, the user also selects a time to let the malware run before any analysis is performed. In relation to rebooting, when a user submits malware, the user may choose to reboot the machine before performing an analysis to kind of trigger the malicious behaviors that the malware might exhibit after rebooting.

[0037] For submissions from honeypots, the program that submits the malware can also submit a set of preferences or parameters for use in the operation of the malware. The exemplary user preferences illustrated in FIG. 5 and the accompanying description are not intended to limit the scope of the present application, but merely to highlight some of the many useful parameters that can be provided in association with one or more potentially malicious files.

[0038] The method 300 also includes optional pre-processing of the submitted file(s) (314). In some implementations, the file is placed into an analysis queue with priority levels (e.g., files submitted with a high priority will be analyzed first). Pre-processing tasks can be performed such as scanning the file with antivirus signatures, determining the file type, detecting packers (i.e., methods to obfuscate the file), and computing cryptographic hashes to determine if the file is already in the database. In some implementations, the system may perform special tasks per file type using “dynamic pre-processors.” As an example, if the file is an Adobe Reader (.pdf), the system will record data on the PDF tags and attempt to extract any hidden JavaScript or other embedded malicious codes. As another example, if the file is a .exe file (Windows executable), the system will perform an analysis of the PE headers and produce alerts (which will show up in the final analysis report) showing which, if any, values indicate malicious behaviors. As yet another example, if a kernel driver is uploaded, the system might perform some tasks that are specific to kernel drivers. In one implementation, a dynamic pre-processor is a Python script that performs a specific task before the primary analysis is performed. Thus, developers do not need to alter the main system source code to use pre-processors—they simply transfer a copy of a Python script in this implementation into a particular directory on the system controller's file system. Then, each time the system receives a file of the specified type, it will automatically pick up the developer's script and execute it.

[0039] The method additionally includes creating a memory baseline (316). The baseline is created as a memory dump for the computer on which the malware is to be operated before the malware is copied to and executed on the computer (e.g., a virtual machine). According to embodiments of the present application, a single baseline may be utilized in conjunction with multiple different malware samples, allowing the user to reuse the initial baseline, rather than requiring the user to create a baseline every single time a malware sample is tested. A virtual machine may be reverted to its original state (i.e., the baseline) after every piece of malware executes. In actual implementations, the system will analyze hundreds to thousands of files. In performing the file analysis, a single baseline can be utilized for one or more of the files to be analyzed.

[0040] The memory baseline includes a copy of all the files that exist on disk before running the malware, along with their complete path to the files, their cryptographic check sums, their sizes, and the like. The memory baseline may also include a copy of the registry files, which contain all the existing registry keys and all the registry values, the status of running processes, threads, and the like. The memory baseline is stored in a database (e.g., MySQL). One of ordinary skill in the art would recognize many variations, modifications, and alternatives.

[0041] The malware file is copied to a machine and executed (318), i.e., the execution system. The machine may be a clean virtual machine or a physical machine as illustrated in FIG. 1. Utilizing the parameters previously specified, the file is executed in the context of the desired operating system in order to determine the effect the file has on the system. The file runs for the desired amount of time as discussed in relation to the user choosing the execution parameters at the time of submission.

[0042] The machine is powered down (320) and the post-execution state of the memory is stored (i.e., a copy of the target system's memory (RAM) for analysis). The machine is powered down in order to access the files on the file system without rootkits and malicious code being able to intercept the ability to enumerate files. As an example, if a rootkit is installed as malware on the system, the rootkit can hide the files when they are accessed for storage. By powering the system down and analyzing the disk from a forensic perspective, the malware detection and analysis system is able to determine a list of the files that exist and their check sums and their sizes without going through the operating system running on the machine.

[0043] The memory baseline and the memory after execution are used to determine the presence of new, missing, and/or modified files. Additionally, embodiments of the present invention collect forensic artifacts from the event log, registry, application log file, and the like. The post-execution data along with the forensic artifacts are stored in a database. Utilizing the obtained artifacts, embodiments are able to generate timelines of activities performed by the malware during operation. Instead of simply reporting on what the malware changed or what it accessed (like conventional systems), embodiments can provide a report including the exact order in which the malware performed its various actions. Examples of forensic artifacts include time stamps that are associated with files, time stamps that are kept for registry keys, time stamps that are stored in the memory dump (i.e., post-execution memory), time stamps located within files on a file system (e.g., application logs or the like), etc.). Thus, time stamps are gathered from a variety of different sources. Analysis of the file system and comparison with the baseline will include an analysis of the time stamps since files with time stamps after the creation of the baseline (i.e., after the time that the file was transferred to and executed on the machine) are associated with files changed either a direct result of the malware or an indirect result of malware. One of ordinary skill in the art would recognize many variations, modifications, and alternatives.

[0044] FIG. 11 is an exemplary simplified diagram illustrating a baseline, a comparison, and a report generated based on the differences between the baseline and comparisons according to an embodiment. Referring to FIG. 11, an example is provided of the different data that can be stored in either the memory baseline or the post-execution memory. If there are ten processes running, the number of processes and the process names, process IDs, and the like can be stored in the memory baseline. The post-execution memory shows that a different number of processes are running, new processes, or the like. Referring to FIG. 11, process svchost.exe (PID 16) was running and is stored in the memory baseline, but is missing from the comparison (i.e., post-execution memory). A new process boo.exe (PID 128) has been initiated by the malware. Similar changes are observed for the file system, the registry, and the like. The examples illustrated in FIG. 11 are not intended to be exhaustive, but merely to highlight some of the information that may be tracked using some embodiments.

[0045] The method 300 further includes analyzing the memory baseline and the memory after execution (324) to determine the changes made by the malware and generating a report (326). The report may include a wide variety of information and can be delivered to the user in a variety of manners. As an example, the report could be provided in pdf format or the user could view the report on the website. The report can either display all of the information in the post-execution memory (e.g., all running processes) or only display the changes with respect to the memory baseline. As illustrated in FIG. 11, the report lists unchanged processes, terminated processes, and new processes, determined by comparing the baseline with the system after running malware. Data that is original, deleted, or new may be illustrated in different colors to emphasize these differences. For example unchanged processes may be displayed in a first color (e.g., gray), terminated processes (i.e., PID 16) in a second color (e.g., red), new processes (i.e., PID 128) in a third color (e.g., green), and modified data in a fourth color (e.g., orange). Other colors may, of course, be implemented or other means of distinguishing the difference in data between the baseline and the modified baseline. The baseline, comparison, and result illustrated in FIG. 11 are only provided as examples, and additional data and analysis can be provided as appropriate to the particular application. For instance, the number of network connections could be utilized in some embodiments. Thus, the memory baseline is not limited to items stored in memory, but can include other aspects of the machine's operation and state. One of ordinary skill in the art would recognize many variations, modifications, and alternatives.

[0046] The report can include a timeline associated with the changes made by the malware. As an example, using the time stamps obtained from a number of sources as discussed above, the times that files are modified and created can be combined with other time stamps related to files on the machine and other time stamps stored in memory to determine a timeline that can be displayed as part of the report. The report can be stored, delivered to the user by various means, including email, posted to one or more websites, printed, and the like.

[0047] Some embodiments distinguish between the state of a machine prior to and after execution of malware based on lists of files, registry keys, and memory objects that exist in the baseline system. After execution of the malware, one or more of the data items in the memory baseline are compared to the post-execution status of the machine. New, missing, and/or modified data in the databases is used as an illustration of this process.

[0048] Examples of information tracked, stored, and compared by some embodiments include, but are not limited to:

[0049] Processes

• • [0050] Process ID
  • [0051] Parent process ID
  • [0052] EPROCESS address
  • [0053] Number of threads
  • [0054] Number of open handles
  • [0055] Full path on disk to the process executable
  • [0056] Process creation time
  • [0057] Process end time (if it exited)
  • [0058] Ssdeep hash of the process image in memory
  • [0059] Command line arguments supplied to the process
  • [0060] YARA (custom antivirus signatures) hits
  • [0061] Process Environment Block address
  • [0062] Boolean values which are 1 (true) if the process exists in two alternate process listings or 0 (false) if the process is not in the alternate process listings
  • [0063] The session in which the process is running
  • [0064] If the process is in the linked list or not

[0065] Window stations

• • [0066] Name of the window station
  • [0067] EPROCESS of member processes

[0068] Desktop objects

• • [0069] Name of the desktop's window station
  • [0070] Name of the desktop
  • [0071] EPROCESS of member processes

[0072] Services

• • [0073] Service Name
  • [0074] Service binary path
  • [0075] Process ID of the running service
  • [0076] State of the service (i.e. stopped, running, paused)
  • [0077] Type of service (i.e. kernel driver, DLL, exe)

[0078] DLLs

• • [0079] Base address of each DLL in a process
  • [0080] Size of each DLL in a process
  • [0081] Path on disk to each DLL
  • [0082] Boolean values which are 1 (true) if the DLL exists in the three module lists or 0 (false) if the DLL does not exist

[0083] Process Memory

• • [0084] VAD Tag for each allocated memory segment in a process
  • [0085] Starting virtual address of allocated memory region
  • [0086] Ending virtual address of allocate memory region
  • [0087] Page protection (i.e. read/write/execute)
  • [0088] Mapped file path
  • [0089] YARA (custom antivirus signatures) hits
  • [0090] Contents of the memory segment if suspicious

[0091] Kernel drivers

• • [0092] Path on disk to the kernel driver loaded
  • [0093] Base address and size of the kernel driver
  • [0094] SSdeep hash of the kernel driver
  • [0095] Entry point of the driver
  • [0096] If the driver is in the linked list or not

[0097] Connections

• • [0098] Owning process
  • [0099] Local IP address
  • [0100] Local Port
  • [0101] Remote IP address
  • [0102] Remote Port
  • [0103] If the connection is linked (hidden)

[0104] Sockets

• • [0105] Owning process
  • [0106] Port
  • [0107] Protocol
  • [0108] Creation time
  • [0109] If the socket is linked (hidden)

[0110] Handles

• • [0111] Owning process
  • [0112] Type of handle (i.e. file, registry key, event)
  • [0113] Object name

[0114] IDT (Interrupt Descriptor Table)

• • [0115] Address of each IDT entry
  • [0116] API name of the IDT handler

[0117] SSDT (System Service Dispatch Table)

• • [0118] Base address of SSDT
  • [0119] Address of each entry
  • [0120] Name of each entry
  • [0121] Owning driver

[0122] Notification routines

• • [0123] Type of notification routine (image load, process create, thread create)
  • [0124] Address of handler
  • [0125] Owning driver

[0126] Threads

• • [0127] Owning process
  • [0128] Thread ID
  • [0129] Starting address of the thread
  • [0130] SSDT used by the thread
  • [0131] Creation time of the thread

[0132] Mutants (mutexes)

• • [0133] Owning process
  • [0134] Owning thread
  • [0135] Name of the mutex

[0136] In addition to the operation of the system as described above, some embodiments enable the utilization of memory forensics capabilities as part of a memory forensics platform. The system described herein can be used to upload memory dumps (i.e., copies of RAM) from potentially compromised machines and the system can be used to extract critical informational, re-construct the state of the system, and produce a report on any malicious activity detected.

[0137] Additionally, some embodiments provide for a rootkit detection component that may include several scripts written in Python or another suitable scripting language. In a particular implementation, the scripts can be used to analyze data in the database for inconsistencies and other behaviors that indicate a rootkit may be present. Because embodiments perform analysis after the machine is powered down, the rootkits are prevented from concealing their presence from the rootkit detection programs. Since embodiments obtain a memory dump and analyze it rather than using the machine's operating system (e.g., Windows), rootkits are detected and reported to the user despite the rootkit's typical ability to protect itself. The memory-based rootkits and behaviors include, but are not limited to:

[0138] Attempts to hide processes

[0139] Attempts to hide kernel drivers

[0140] Attempts to inject code or unlink DLLs

[0141] Attempts to hide network sockets and connections

[0142] Attempts to hide suspicious data in process memory

[0143] Attempts to install hook in user or kernel mode memory (IAT, EAT, Inline)

[0144] Attempts to modify the IDT, SSDT, and driver IRP entries

[0145] Attempts to hide threads

[0146] Attempts to register malicious notification routines

[0147] Attempts to install promiscuous mode packet sniffers

[0148] File system-based rootkits and behaviors can include, but are not limited to:

[0149] Attempts to change timestamps on files

[0150] Attempts to disguise files by using an invalid file extension

[0151] Attempts to patch or add code to legitimate system binaries

[0152] Attempts to modify the master boot record (MBR)

[0153] Attempts to hide in unallocated/slack disk space

[0154] Attempts to hide in NTFS ADS (alternate data streams)

[0155] Attempts to disable Windows File Protection

[0156] Attempts to spread via autorun.inf files

[0157] Registry-based rootkits and behaviors can include, but are not limited to:

[0158] Attempts to disable the Windows host firewall

[0159] Attempts to start automatically with registry run keys

[0160] Attempts to overwrite or add new cryptography certificates to the system

[0161] Attempts to disable Windows File Protection (WFP)

[0162] Attempts to disable safe boot

[0163] Attempts to disable security services and antivirus products

[0164] Attempts to alter browser behaviors and plug-ins

[0165] Some embodiments track relationships between files and malware samples. As an example, if a user submits a suspicious zip file, the system may open the zip file and extract the zipped files (e.g., three executable files) and automatically tag them as coming from the same zip file. Thus, the malware detection and analysis system retains information relating these three executable files to each other through the zip file although the files will be analyzed individually. Continuing with the example, the first file is analyzed in the virtual machine, the virtual machine can be reverted, the second file is analyzed in the virtual machine, the virtual machine can be reverted a second time, and the third file is analyzed in the virtual machine. The three files are linked together in the database so that at a later point in time the user can relate the files to each other. Searching on one of the files will enable the user to locate the related files that were included in the zip file.

[0166] Some embodiments may analyze malware files that operate in conjunction with each other, although independent operation may not result in malicious behavior. Additionally, multiple files can be analyzed, for example, the three files in the zip file discussed above, while the other files are existing in place. Additionally, the files can be placed in specific folders in the operating system and operation as a function of the folder location may be analyzed. For example, a piece of malware may check the directory in which it is stored. If it not located in the C:\Windows\System32 directory, the malware will not produce any result. If the malware is placed in the C:\Windows\System32 directory, then it will produce a malicious result. Thus, embodiments enable the placement of one or more files, which may be related, into folders as desired by the user.

[0167] Additionally, embodiments may track files using a tree structure. As an example, a malware sample is run, drops two other files, then contacts some site on the Internet and downloads five other files. The five new files will be detected during the method illustrated in FIG. 3 and then submitted for analysis individually. The tree structure develops as the five new files, as they are executed during analysis, can drop their own files and download other files off the Internet. Each file then is treated as a node, with the tree structure showing the relationship between parent and child files. The relationship between files can be tracked and reported out.

[0168] Embodiments may include a reverse engineering integrated development environment (IDE) based on the flexible nature of the system described herein. Thus, a reverse engineer is provided with understanding and the ability to control the flow of execution of each malware sample at a very low level. Additionally the reverse engineer is able to use the system to decode, decrypt, or unpack malware samples if the malware authors have implemented such protections.

[0169] Embodiments may support “dynamic debugger plugins,” which are Python scripts that execute a malware sample under the context of a debugger. A debugger script can start and stop the malware from executing at any time, analyze the contents of the CPU registers, or change the values of variables or strings within the malware's memory in order to alter its behaviors. One example of such a plugin patches instructions in a malware sample that make it execute on systems that it otherwise would not infect. For example, some malware samples are locked to specific hardware—attackers do this to specifically prevent sandbox analysis.

[0170] FIG. 4 is an exemplary simplified diagram illustrating operation of malware under control of a debugger according to an embodiment. As illustrated in FIG. 4, the image on the left portion of the figure shows how malware operates on an “incorrect” system. As an example, a Trojan could be hardware locked—the Trojan uses some of the hardware of the infected machine (e.g., the serial number of the hard drive) to encrypt and store information on the hard drive particular to the infected machine. If the Trojan is then copied onto an analysis system, it will not find the needed serial number and will not perform malicious behavior. The image on the right side of the figure shows how the same malware operates on the same system, but using a debugger plugin to patch the program's logic, thereby bypassing the malware's limitation and allowing the system to execute the malware.

[0171] Additionally, embodiments may provide a reverse engineering IDE that streamlines the process of unpacking and decrypting data. As described below, embodiments provide modules that support algorithms and functions that conventional systems do not support. If a malware sample is packed, one of the pre-processors discussed in relation to step 314 of FIG. 3 can determine the type of packer used on the sample and attempt to automatically unpack it. The unpacking procedure consists of emulators, command-line programs in Linux, dumping the malware from memory, and the like. If the system is able to unpack the malware, a copy of the unpacked malware is associated with the original sample in the database.

[0172] A large number of malware samples use well-known algorithms such as base64, ROR, ROT13, and XOR to encode files and data that they store on an infected machine's disk. When the system performs the file system analysis and detects new files that may be encoded using one of the well-known algorithms, it automatically decodes the file. This instantly reveals configurations, command and control protocols, and stolen information that attackers try to hide. Users are then able to view the original (encoded) or the modified (decoded, plain text) files from the web site. Additionally, some embodiments are pre-configured with several rare algorithms implemented by specific malware families (i.e. not just well-known packing algorithms).

[0173] Embodiments may utilize a number of database component and may run MySQL. The main tables that may be included in the database include, but are not limited to:

• • [0174] The “submissions” table may contain one column for each of the preferences associated with a malware sample (for example, which version of Windows to execute it on, how long to let it run, and so on).
  • [0175] The “binaries” table may contain the malware sample itself, along with the MD5, file type, file size.
  • [0176] The “groups” table may associate IDs from the submissions table when multiple submissions are part of the same security incident or related in some other manner.
  • [0177] The “avscan” table may store the list of antivirus detections for each malware sample and any files that it drops or downloads that were detected by an antivirus product.
  • [0178] The “pefiles” tables may store artifacts of PE headers such as PE section names, entropy for each section, file characteristics, and imported/exported API functions.
  • [0179] The “network” table may store data on traffic generated by each malware sample. It includes source and destination IP address, ports, protocols, and packet payloads (the data transmitted).
  • [0180] The “files” table may store information on files created, modified, or deleted during an analysis. It stores timestamps of all file system related activity.
  • [0181] The “registry” table may store information on the registry keys or values that malware created, modified, or deleted.
  • [0182] The “events” table may store event log entries that were logged on the target system while the malware was running.
  • [0183] The “behaviors” table may contain a lengthy collection of different malicious behaviors that the malware detection and analysis system can detect. Each malware sample has a row in the behavior table, which describes its functionality set. Examples of behaviors that can be stored in this table are listed in the table's columns and a description of each is provided:

DROP TABLE IF EXISTS ‘am_behaviors‘;
CREATE TABLE ‘am behaviors‘ (
‘id‘ int(11)	NOT NULL,
‘ads‘ tinyint(1)	DEFAULT 0,	-- Hides in NTFS ADS streams
‘mbr‘ tinyint(1)	DEFAULT 0,	-- Modifies MBR
‘rawsock‘ tinyint(1)	DEFAULT 0,	-- Promiscuous mode NIC
‘bho‘ tinyint(1)	DEFAULT 0,	-- Installs BHO
‘newsvc‘ tinyint(1)	DEFAULT 0,	-- Creates services
‘disablesec‘ tinyint(1)	DEFAULT 0,	-- Disablesecurity/av services
‘apihook‘ tinyint(1)	DEFAULT 0,	-- Hooks API functions
‘inject‘ tinyint(1)	DEFAULT 0,	-- Injects code to other processes
‘drv‘ tinyint(1)	DEFAULT 0,	-- Loads a driver
‘autostart‘ tinyint(1)	DEFAULT 0,	-- Autostart with Windows
‘disablefw‘ tinyint(1)	DEFAULT 0,	-- Disables firewall
‘listensock‘ tinyint(1)	DEFAULT 0,	-- Opens a listening socket
‘wrsysdir‘ tinyint(1)	DEFAULT 0,	-- Writes to the SYSTEM32 directory
‘dlexe‘ tinyint(1)	DEFAULT 0,	-- Downloads an EXEIDLLISYS
‘ids‘ tinyint(1)	DEFAULT 0,	-- Trips Snort IDS alert(s)
‘banker‘ tinyint(1)	DEFAULT 0,	-- Banker / password stealer
‘globaldll‘ tinyint(1)	DEFAULT 0,	-- Maps DLL from \\?\Globalroot\
‘globalnotify‘ tinyint(1)	DEFAULT 0,	-- Sets thread, proc, or image load notify routine
‘wipesr‘ tinyint(1)	DEFAULT 0,	-- Deletes system restore points
‘importdna‘ int(11)	DEFAULT 0,	-- DNA malicious index based on IAT functions (see
		am_alerts)
‘hookirp‘ tinyint(1)	DEFAULT 0,	-- Hooks driver IRP functions
‘ssdt‘ tinyint(1)	DEFAULT 0,	-- Modifies SSDT (hooks or makes copies)
‘timestamp‘ tinyint(1)	DEFAULT 0,	-- Sets file creation times to past dates
‘misidentify‘ tinyint(1)	DEFAULT 0,	-- Creates files with mis-matched extensions
‘autorun‘ tinyint(1)	DEFAULT 0,	-- Spreads through autorun
‘brutepw‘ tinyint(1)	DEFAULT 0,	-- Attempts to brute force passwords
‘movefileex‘ tinyint(1)	DEFAULT 0,	-- Deletes files with MoveFileEx
‘safeboot‘ tinyint(1)	DEFAULT 0,	-- Alters safe boot configuration
‘regexec‘ tinyint(1)	DEFAULT 0,	-- Writes values with exe, sys, dll extensions to
		registry
‘browser‘ tinyint(1)	DEFAULT 0,	-- Modifies browser settings
‘tasks‘ tinyint(1)	DEFAULT 0,	-- Creates scheduled tasks/jobs
‘spam‘ tinyint(1)	DEFAULT 0,	-- Sends emails including but not limited to spam
‘wfp‘ tinyint(1)	DEFAULT 0,	-- Defeats Windows file protection
‘unlinkdll‘ tinyint(1)	DEFAULT 0,	-- Unlinks DLLs
‘netmod‘ tinyint(1)	DEFAULT 0,	-- Modifies network/DNS settings
‘vmdetect‘ tinyint(1)	DEFAULT 0,	-- Detects VMs or other monitors
‘encode‘ tinyint(1)	DEFAULT 0,	-- Encodes/encrypts/compresses data
‘caroot‘ tinyint(1)	DEFAULT 0,	-- Adds new root CA certs
INDEX (id)
) ENGINE=MyISAM;

[0184] Embodiments may provide an environment (also referred to as a sandbox) that leverages advanced memory forensics to report on the state of a computer after the computer is infected with malware. Analysts are provided with greater control over the target environment, greater control of the type of information gathered during analysis, and the ability to go deeper into rootkit detection than by using conventional systems. Embodiments also include analysis and reporting on information such as changes to the file system, registry, and network.

[0185] Referring to FIG. 5, some embodiments provide a user interface through which a user may submit files and choose analysis preferences or parameters. In addition to files, a user can enter a URL. Multiple file types are accepted, including PDF files and multiple types of PE files: for example, EXEs, DLLs, and kernel drivers. Submissions are also accepted by using a script as well as by email attachments, links in email bodies, tweets, and the like.

[0186] FIG. 6 is an exemplary display screen of a report dashboard according to an embodiment. As illustrated in the embodiment in FIG. 6, each submitted file has its own web page that starts off with a dashboard full of useful information. From here, a user can download the Automatic Cleaner Script (.acs), find out if the malware (or any component thereof) sniffs network traffic, injects code into other processes, and things of that nature. A user can download the post-infection registry hives, a full packet capture, or the submitted file (in case the user needs to recover it).

[0187] FIG. 7 is an exemplary display screen illustrating a screen showing different anti-virus vendors' assigned names for a particular malware implementation. If the user submits the malware code to be analyzed by VirusTotal, a well-known malware scanner, the results will be displayed on this page. By clicking the virus name, the user can either show any other samples in the database with a matching name or go directly to an Internet search of the name.

[0188] Table 1 is an example of an events log analysis. This table shows any changes to the Windows event logs (System, Application, and Security) and highlights them if they appear suspicious. In this case, the malware started the “Net Login” service and terminated two others related to security (the security center and a component of the firewall). Additional columns, such as Record ID and Event ID, may be available.

TABLE 1
Log	Source	Type	Time	Message
Application	SecurityCenter	EVENTLOG_INFORMATION_TYPE	1256158881
Application	ESENT	EVENTLOG_INFORMATION_TYPE	1251840039	wuaueng.dll 460
				SUS20ClientDataStore: 0
Application	ESENT	EVENTLOG_INFORMATION_TYPE	1251840039	wuauelt 460 5 01 2600
				2180
Application	ESENT	EVENTLOG_INFORMATION_TYPE	1251477798	wuaelt 1532
Application	ESENT	EVENTLOG_INFORMATION_TYPE	1251477798	wuaueng.dll 1532
				SUS20ClientDataStore: 0
System	EventLog	EVENTLOG_INFORMATION_TYPE	1256159175
System	Service Control	EVENTLOG_INFORMATION_TYPE	1256159174	Net Login running
	Manager
System	Service Control	EVENTLOG_INFORMATION_TYPE	1256159174	Net Login start
	Manager
System	Service Control	EVENTLOG_INFORMATION_TYPE	1256158882	Security Center stopped
	Manager
System	Service Control	EVENTLOG_INFORMATION_TYPE	1256158881	Application Layer
	Manager			Gateway Service stopped

[0189] Table 2 is an example of a registry analysis. The registry analysis page shows created keys/values in green, modified entries in orange and deleted entries in red. This table shows the entries that the malware created in order to start the “Net Login” service mentioned above.

TABLE 2
HKEY_LOCAL_MACHINE\SYSTEM\ControlSet001\Services\wsesvc
Start	REG_DWORD	0x00000002 (2)	0x00000004 (4)
HKEY_LOCAL_MACHINE\SYSTEM\ControlSet001\Services\SharedAccess
Start	REG_DWORD	0x00000002 (2)	0x00000004 (4)
HKEY_LOCAL_MACHINE\SYSTEM\ControlSet001\Services\NetLogin
Start	REG_DWORD		0x00000002 (2)
Type	REG_DWORD		0x00000010 (16)
DisplayName	REG_SZ		Net Login
ImagePath	REG_EXPAND_SZ		C:\WINDOWS\svchost exe
ObjectName	REG_SZ		LocalSystem
ErrorControl	REG_DWORD		0x00000001 (1)

[0190] Table 3 is an example of a file system analysis. This table shows the files that the malware created or deleted. If the user chooses a “deep” scan on the submission page instead of a “quick” scan (which is the default in some embodiments), then it also detects changes to files based on MD5/SHA1 hash. It shows a preview of the file's contents and the file's type, so the user can easily tell if ssv.txt is really an MS-DOS executable and see that Igate.htm probably contains some encoded command and control data. Based on the content type, the user can get extended information, as shown in the next section (for example PE header details for MS-DOS programs or extracted JavaScript from malicious PDF files).

TABLE 3
created	C:\Documents and Settings\miles\Local Settings\Temporary Internet Files\Content	152	ASCII text, with no line
	IE5\UBWDANOL\lgate[1].htm		terminators
0000	4d	43	42	6f	64	48	52	77	4f	69	38	76	63	32	78	6c	MCBodHRw	0i8wc2x1
0010	5a	58	42	68	64	47	35	70	5a	32	58	30	4c	6d	4e	75	ZXBhdG5p	Z2h0LmNu
0020	4c	32	4a	30	4e	79	39	6d	62	33	56	30	4c	6e	42	6f	L2J0Ny9m	B3V0LnBo
0030	63	43	41	78	49	47	68	30	64	48	41	36	4c	79	39	7a	cCAxIGh0	dHA6Ly9z
0040	62	47	56	6c	63	47	46	30	62	6d	6c	6e	61	48	51	75	bGV1cGF0	bmlnaHQu
0050	59	32	34	76	62	47	6c	69	4c	32	39	7a	62	59	35	30	Y24vbGli	L29zbi50
0060	65	48	51	67	4d	53	42	6f	64	48	52	77	4f	69	38	76	eHQgMSBo	dHRw0iSv
0070	63	32	78	6c	5a	58	42	68	64	47	35	70	5a	32	58	30	c2x1ZXBh	dG5pZ2h0
created	C:\Documents and Settings\miles\Local Settings\Temporary Internet Files\Content	88576	MS-DOS executable PE
	IE5\UBWDANOL\ssv[1].txt		for MS Windows (GUI)
			Intel 80386 32-bit
0000	4d	5a	90	00	03	00	00	00	04	00	00	00	ff	ff	00	00	MZ......	........
0010	b8	00	00	00	00	00	00	00	40	00	00	00	00	00	00	00	........	@.......
0020	00	00	00	00	00	00	00	00	00	00	00	00	00	00	00	00	........	........
0030	00	00	00	00	00	00	00	00	00	00	00	00	80	00	00	00	........	........
0040	0e	lf	ba	0e	00	b4	09	cd	21	b8	01	4c	cd	21	54	68	........	!..L.!Th
0050	69	73	20	70	72	6f	67	72	61	6d	20	63	61	6e	6e	6f	is progr	am canno
0060	74	20	62	65	20	72	75	6e	20	69	6e	20	44	4f	53	20	t be run	in DOS
0070	6d	6f	64	65	2e	0d	0d	0a	24	00	00	00	00	00	00	00	mode....	$.......

[0191] Table 4 is an example of file information (for PE files). This report is available for any PE files in the database, including the submission itself, any PE files that the submission drops, any PE files extracted from network traffic, and any EXE, DLL, or SYS file extracted out of memory. The user can view the file's meta data or check over critical PE header values (highlighted values indicate packing). The second part of the page (not shown) shows imported and exported functions, any visible strings in the binary, and a hex dump.

TABLE 4
File Meta Data
Size	0x15a00
MD5	1c5e79f5f4caab5f5c9a69ab91d478b2
SHA1	428d52728c29ec557fle4df282ab76af70230823
SSDEEP	1536 7g+Ohs17i2uGSuoWi9VVBwwbDHjM5EeK814JvW8RPkSi/CEKCH]
PE Header
	ImageBase	0x400000
	SizeOfImage	0x37000
	NumberOfSections	0x4
	AddressOfEntryPoint	0x10f5 (text)
	Characteristics	0x10f
	TLS Directory	1
	Resource Directory	1
	TimeDateStamp	0x496c1d06 (Tue Jan 13 4:48:06 UTC 2009)
PE Sections
	Section Name	VirtualAddress	VirtualSize	RawSize	Entropy
	.rsrc	0x17000	0x20000	0x1200	3.152260
	.rdata	0x8000	0x3fa	0x400	5.011950
	.data	0x9000	0xdba6	0xdc00	6.770210
	.text	0x1000	0x62e8	0x6400	7.121550

[0192] Tables 5, 6, and 7 are examples of network traffic analysis. This table shows information on communications with outside hosts. It shows any tripped Snort IDS alerts, a port and protocol summary, extended details of HTTP requests, GeoIP plot-mapping (not shown), and a tool for visual breakdown of traffic into protocols, ports, and sessions.

TABLE 5
Traffic Log (Anything Not WWW/DNS)
			Source			Destination
Protocol	Source IP	Co.	Port	Destination IP	Co.	Port
TCP	172.16.99.130		1104	218.93.205.30	China	65520
UDP	172.16.99.130		1112	239.255.255.250		1900
TCP	172.16.99.130		1118	202.97.184.196	China	81
UDP	172.16.99.130		1112	239.255.255.250		1900
UDP	172.16.99.130		1112	239.255.255.250		1900
TCP	172.16.99.130		1122	202.97.184.196	China	81
DNS Traffic
Query	Type	Class	Domain	Value
Query	A	IN	proxim.ircgalaxy.p1
Query	A	IN	time.windows.com
Queyr	A	IN	time.windows.com
Answer	CNAME	IN	time.windows.com	time.microsoft.akadns.net
Answer	A	IN	time.microsoft.akadns.net	207.46.232.182
Query	A	IN	proxim.ircgalaxy.p1
Answer	A	IN	proxim.ircgalaxy.p1	218.93.205.30

TABLE 6
Connection To: GET sleepatnight.cn/op/lgate.php?n=46E3956E5CEA462A (91.206.201.39.80)
GET /op/lgate.php?n=46E3956E5CEA462A HTTP/1.0
Accept: */*
User-Agent: Mozilla/4.0 (compatible, MSIE 6.0, Windows NT 5.1; SV1)
Host: sleepatnight.cn
Connection: Keep-Alive
0000	47	45	54	20	2f	62	74	37	2f	66	6f	75	74	2e	70	68	GET /bt7	/fout.ph
0010	70	20	48	54	54	50	2f	31	2e	30	0d	0a	41	63	63	65	p HTTP/1	.0..Acce
0020	70	74	3a	20	2a	2f	2a	0d	0a	55	73	65	72	2d	41	67	pt: */*.	.User-Ag
0030	65	6e	74	3a	20	4d	6f	7a	69	6c	6c	61	2f	34	2e	30	ent: Moz	illa/4.0
0040	20	28	63	6f	6d	70	61	74	69	62	6c	65	3b	20	4d	53	(compat	ible; MS
0050	49	45	20	36	2e	30	3b	20	57	69	6e	64	6f	77	73	20	IE 6.0;	Windows
0060	4e	54	20	35	2e	31	3b	20	53	56	31	29	0d	0a	48	6f	NT 5.1;	SV1)..Ho
0070	73	74	3a	20	73	6c	65	65	70	61	74	6e	69	67	68	74	St: slee	patnight
Response Code: 200, Content Type: data, Length: 135469 (View Details)
HTTP/1.1 200 OK
Server: nginx
Date: Wed, 21 Oct 2009 20:59:37 GMT
Content- Type: text/html
Connection: keep-alive
X-Powered-By: PHP/5.2.11
Content-Length: 152
0000	4d	43	42	6f	64	48	52	77	4f	69	38	76	63	32	78	6c	MCBodHRw	0i8vc2xl
0010	5a	58	42	68	64	47	35	70	5a	32	68	30	4c	6d	4e	75	ZXBhdG5p	Z2h0LmNu
0020	4c	32	4a	30	4e	79	39	6d	62	33	56	30	4c	6e	42	6f	L2J0Ny9m	B3V0LnBo
0030	63	43	41	48	49	47	68	30	64	48	41	36	4c	79	39	7a	cCAxIGh0	dHA6Ly9z
0040	62	47	56	6c	63	47	46	30	62	6d	6c	6e	61	48	41	75	bGV1cGF0	bmlnaHQu
0050	59	32	34	76	62	47	6c	69	4c	32	39	7a	62	69	35	30	Y24vbHli	L29zbi50
0060	65	48	51	67	4d	53	42	6f	64	48	52	77	4f	69	38	76	eHQgMSBo	dHRw0i8v
0070	63	32	78	6c	5a	58	42	68	64	47	35	70	5a	32	68	30	C2xlZXBh	dG5pZ2h0

TABLE 7
TCP/UDP/. . . Sessions
1.	Wed. Oct 21	0 s	172.16.99.254 −>	ICMP	20 bytes	Echo
	16:01:04 2009		172.16.99.130			raw
2.	Wed. Oct 21	77 s	172.16.99.130:1085 <−>	domain	1133 bytes	raw raw
	16:01:06 2009		172.16.99.2:53			as html
3.	Wed. Oct 21	12 s	172.16.99.130:1101 <−>	domain	742 bytes	raw raw
	16:01:07 2009		172.16.99.2:53			as html
4.	Wed. Oct 21	97 s	172.16.99.130:1104 −>	65520	312 bytes	raw raw
	16:01:07 2009		218.93.205.30:65520
5.	Wed. Oct 21	1 s	172.16.99.130:1107 −>	www	13645 bytes	raw raw
	16:01:16 2009		91.206.201.39:80			as html
						session bytes

[0193] Table 8 is an example of a memory forensics/analysis. Embodiments may include strong memory forensics capabilities. As described throughout the present specification, some embodiments may determine new processes, hidden or injected code, loaded kernel modules, network connections, open files, API hooks, active service processes, and the like. Embodiments also include a built-in custom memory scanner that is quick and configurable. Table 8 shows some positive hits in the memory of wuauclt.exe and winlogon.exe.

TABLE 8
0x1cc (wuauclt.exe)	VadS	0x02ad0000	0x02adfff	0x6		0x0
0x1cc (wuauclt.exe)	Vad	0x7ffa0000	0x7ffa6fff	0x6	memory.460.7ffa0000-7ffa6fff.dmp	0x0
0x69c (winlogin.exe)	Vad	0x00400000	0x00431fff	0x7	memory.1692.400000-431fff.dmp	0x4
					0x1d42e => IE: Password-Protected
					0x1d860 => paypal.com
					0x1d42e => IE: Password-Protected
					0x1d860 => paypal.com
0x69c (winlogin.exe)	VadS	0x003b0000	0x003b0fff	0x6		0x0
0x69c (winlogin.exe)	VadS	0x003a0000	0x003a4fff	0x6	memory.1692. 3a0000-3afff.dmp	0x1
					0x6bb => &pass
0x69c (winlogin.exe)	VadS	0x003c0000	0x003c0fff	0x6		0x0
0x69c (winlogin.exe)	VadS	0x003d0000	0x003d0fff	0x6		0x0
0x69c (winlogin.exe)	VadS	0x003e0000	0x003e0fff	0x6		0x0
0x69c (winlogin.exe)	VadS	0x003f0000	0x003f0fff	0x6		0x0
0x69c (winlogin.exe)	VadS	0x00d40000	0x00d40fff	0x6		0x0

[0194] Table 9 shows which API functions the malware hooked inside the explorer.exe process. It also shows the type of hook (INLINE), the relevant memory addresses, and disassembled instructions.

TABLE 9
0x644 (explorer.exe)	INLINE ntdll. dll@ 0x7c900000	NtCreatedFile	0x7c90d682	0x7ffa2394	call	0x7ffa2394
0x644 (explorer.exe)	INLINE ntdll. dll@ 0x7c900000	NtCreatedProcess	0x7c90d754	0x7ffa23e8	call	0x7ffa23e8
0x644 (explorer.exe)	INLINE ntdll. dll@ 0x7c900000	NtCreatedProcessEx	0x7c90d769	0x7ffa2315	call	0x7ffa23f5
0x644 (explorer.exe)	INLINE ntdll. dll@ 0x7c900000	NtOpenFile	0x7c90dcfd	0x7ffa23el	call	0x7ffa23el
0x644 (explorer.exe)	INLINE ntdll. dll@ 0x7c900000	ZwCreatedFile	0x7c90d682	0x7ffa2394	call	0x7ffa2394
0x644 (explorer.exe)	INLINE ntdll. dll@ 0x7c900000	ZwCreatedProcess	0x7c90d754	0x7ffa2e84	call	0x7ffa2e84
0x644 (explorer.exe)	INLINE ntdll. dll@ 0x7c900000	ZwCreatedProcessEx	0x7c90d769	0x7ffa23f5	call	0x7ffa23f5
0x644 (explorer.exe)	INLINE ntdll. dll@ 0x7c900000	ZwOpenFile	0x7c90dcfd	0x7ffa23el	call	0x7ffa23el
Running Processes (Top)
Name	Pid	EPROCESS	Threads	Handles	Created	Ended	Cmdline
System	0x4	0x1bcc9c8	0x37	0x107		0
smss.exe	0x214	0x1ad3d38	0x3	0x15	1251417475	0	\SystemRoot\System32\smss.exe
csrss.exe	0x25c	0x1aea020	0xb	0x166	1251417483	0	C:\WINDOWS\system32\csrss.cx
							Windows=On SubSystemType=W
							ServerDll=winsrv:UserServerD1lln
							ProfileControl=Off MaxRequestTh
winlogon.exe	0x274	0x1b33a48	0x13	0x226	1251417483	0	winlogin.exe
services.exe	0x2a0	0x1986020	0x11	0x102	1251417483	0	C:\WINDOWS\system32\csrss
lsass.exe	0x2ac	0x1aaf020	0x18	0x14f	1251417483	0	C:\WINDOWS\system32\csrss
vmacthlp.exe	0x348	0x1970550	0x2	0x1f	1251417483	0	“C&P:\Program Files\VMware\VMwa
svchost.exe	0x354	0x19726e8	0x13	0xc9	1251417484	0	C:\WINDOWS\system32\svchost
svchost.exe	0x3a4	0x1974da0	0xc	0xe2	1251417484	0	C:\WINDOWS\system32\svchost
svchost.exe	0x400	0x1ae458	0x37	0x409	1251417484	0	C:\WINDOWS\system32\svchost
svchost.exe	0x42c	0x1abbda0	0x7	0x59	1251417484	0	C:\WINDOWS\system32\svchost
svchost.exe	0x458	0x1a9ada0	0x11	0xd7	1251417484	0	C:\WINDOWS\system32\svchost
spoolsv.exe	0x5c8	0x1af7da0	0xb	0x79	1251417485	0	C:\WINDOWS\system32\spoolsv
explorer.exe	0x644	0x1770a98	0xb	0x140	1251417486	0	C:\WINDOWS\Explorer.EXE
WMwareTray.exe	0x684	0x1b24878	0x2	0x23	1251417486	0	“C&P:\Program Files\VMware\VMwa
VMwareUser.exe	0x68c	0x1a62020	0x5	0x4c	1251417486	0	“C&P:\Program Files\VMware\VMwa
VMwareService.e	0x77c	0x164ab28	0x5	0xaa	1251417494	0	“C&P:\Program Files\VMware\VMwa
wuauclt.exe	0x1cc	0x1aa8da0	0x8	0xb7	1251840039	0	C:\WINDOWS\system32\wuauc1l
							[400] SUSDS8e4afe12626a5f40b1
winlogin.exe	0x69c	0x1aaa228	0x2	0x27	1256158873	0	C:\WINDOWS\system32\winlogo.
VRT1.tmp	0x70	0x16383a0	0x0	0xffffffff	1256158876	. . . 891
svchost.exe	0x7cc	0x16ac5c8	0x8	0x7a	1256158878	0	svchost.exe C&P:\WINDOWS\TEMP
4.tmp	0x540	0x1afbc70	0x0	0xffffffff	1256158880	. . . 881
9129837.exe	0x600	0x16a2da0	0xb	0x9f	1256158881	0	“C:\WINDOWS\9129837.exe”
cmd.exe	0x1a0	0x1631020	0x0	0xffffffff	1256158881	. . . 882
5.tmp	0x694	0x1766da0	0x0	0xffffffff	1256158881	. . . 895
isvchost.exe	0x380	0x1afdda0	0x5	0x87	1256158895	0	“C:\WINDOWS\isvchost.exe”

[0195] FIG. 8 is an exemplary display screen illustrating searching and reporting functions according to an embodiment. As illustrated in FIG. 8, the user can search for malware by hash, a/v name, a file it created, a host it contacted, an API function that it imports or exports, and a number of other data types. The user can group multiple pieces of malware into a single incident (for example, if a user finds 4 EXEs and 1 DLL on a single system, the user can upload them all and then group them together) and can bookmark reports if the user is too busy, and then go back and review them later.

[0196] Some of the existing reports included as part of the searching and reporting function (not shown in FIG. 8) include, but are not limited to:

[0197] Most recent submissions

[0198] Most commonly hooked API functions

[0199] Most recently API-hooking malware

[0200] Most recent dropped files (from any report)

[0201] Most recently accessed URLs (from any report)

[0202] Most TCP/UDP-traffic producing malware

[0203] Recent malware with low/no anti-virus detection

[0204] Most common packers

[0205] Most common malware families

[0206] Most common imported API functions

[0207] Embodiments provide methods and systems to automate malware detection and analysis. Embodiments compliment existing infrastructure and provide an entirely new set of characteristics. Therefore, products offered by the present assignee are provided with a greater capacity for dynamically analyzing malware under a broader range of conditions. For example, some of the new options described herein include the ability to run malicious code with different privilege levels or on systems with different service packs installed.

[0208] Embodiments provide systems (also referred to as a malware sandbox) based largely or entirely on forensic data acquisition. Conventional sandboxes detect changes to the system by intercepting (or hooking) Win32 and/or kernel-level API calls. Unfortunately, in order to provide comprehensive monitoring, this conventional method creates a tainted analysis station with several hundred (or more) hooks in the operating system. Not only is it very simple to detect hooks, but it is possible for malware to install separate hooks which disable the original hooks and render the analysis system completely ineffective.

[0209] Some embodiments, on the other hand, detect changes to the system by forensically analyzing the contents of physical memory (RAM) after running the malicious binary. Embodiments can detect changes to the system in astonishing new ways, without running special monitoring tools on the analysis station that could potentially cause interference. Additionally, the systems described herein shut down the analysis system and mount the hard drive read-only in order to extract dropped files. This enables the recovery of critical data which is normally not accessible on live systems infected with Trojans such as Tigger, which hooks NTFS and FAT file system drivers.

[0210] Embodiments may provide a web front-end for malware submissions. In an embodiment, the web form generates a queue file for each submission, which contains the user-selected options. A daemon process watches for the creation of new queue files, and picks them up immediately for processing. The framework may utilize VMware's VIX API in an embodiment, which lets a processing script on the VM host control various operations of the VM guests. For example, the processing script using the VIX API can transfer submitted files to a VM guest and then execute it in the VM guest.

[0211] FIG. 9 is an exemplary display screen illustrating another graphical user interface according to an embodiment. Referring to FIG. 9, the web front-end and configuration options are illustrated for one embodiment. Choose a file to upload: This option specifies the path on a user's hard disk to the file to upload/submit. Sleep time: This option specifies the amount of time (in seconds) that the submitted file should run before the malware detection and analysis system begins analyzing the system for changes. If the default of twenty seconds is not long enough for the malware to complete execution, it is possible to re-queue the file for a longer period. Reboot before analysis: This option specifies if the infected system should reboot before the malware detection and analysis system begins analyzing the system for changes. In several cases, the payload of a Trojan might not activate until the first reboot. For example, a Trojan dropper might enter itself into the registry to automatically start with Windows and then terminate. Additionally, payloads such as the kernel driver dropped by the Torpig MBR rootkit do not take effect until the first reboot. This option to reboot ensures greater flexibility to deal with different scenarios. Post-execution program: This option specifies the path to a program on the analysis station that should execute after the malware, but before the malware detection and analysis system analyzes the system for changes. This option is similar to the reboot option; however it aims to satisfy the payloads that only activate on start-up of particular programs, such as Internet Explorer. For example, a malicious Browser Helper Object (BHO) would not activate immediately after a Trojan drops it, nor on first reboot of the system. The BHO would only activate when a new instance of Windows Explorer or Internet Explorer started up. Notify email: This option specifies the email address to which the malware detection and analysis system will send a notification upon completing the analysis. Guest OS: This option specifies the guest operating system in which to run the suspect binary. The options include XP service pack 2, XP service pack 3, Vista, and other suitable operating systems. This feature allows the user to determine exactly which operating systems (and patch levels) are affected by a particular malware or exploit. For example, it is possible to submit a binary to both XPSP2 and XPSP3 and figure out which functionality of the malware succeeds and fails on one or both of the platforms. This information is useful to administrators who need to quickly pinpoint which configurations in their environment are vulnerable. Run as administrator: This option specifies the privilege level at which the malicious code should execute. Users routinely want information regarding which privilege levels are necessary for malware features to properly execute. Similar to the information provided by the “Guest OS” feature, the information provided by the “Run as administrator” feature provides researchers with the necessary knowledge to protect their corporate networks. Monitor API: This option enables usermode API monitoring of the suspect binary. This option provides a detailed trace of API usage at the expense of eliminating the forensic soundness of the analysis station. By enabling the API monitor, the malware detection and analysis system injects DLLs into the suspect binary as it executes. Capture screen: This option specifies if the malware detection and analysis system should capture screen shots of the analysis system's desktop during execution of the malware. If the malware produces visible windows by spawning other processes or creating a window of its own, this option allows researchers to view those windows. Comment: This option is simply to help users track their submissions. For example, if a user submits the same file multiple times (to run as admin, non-admin, and with or without API monitoring), they might want to leave a note to distinguish between the analysis results. Other embodiments provide more, fewer, or different options and preferences. One of ordinary skill in the art would recognize many variations, modifications, and alternatives.

[0212] By analyzing the contents of physical memory, embodiments can quickly determine changes to the system that resulted from running the malicious file. The list of capabilities includes, without limitation:

[0213] New running processes: Embodiments detect new processes without relying on Win32 API (or kernel API) functions to enumerate the list. Therefore, providing defense against rootkits that intentionally try to hide running processes. By identifying EPROCESS structures in memory, the malware detection and analysis system can detect even the stealthiest rootkits that hide using DKOM (direct kernel object manipulation). Since EPROCESS structures persist in memory even after the process exists, it is also possible to determine processes which ran, but then terminated before analysis.

[0214] Table 10 shows the newly created process names, their respective process IDs, and the command-line parameters (if any).

TABLE 10
Processes:
+certproc32.exe	1740	“C:\WINDOWS\certproc32.exe” [ ]
+cmd.exe	1852	“C:\WINDOWS\system32\CMD.EXE” /C del C:\297681~1.EXE>> NUL [ ]

[0215] New DLLs: Embodiments may detect new DLLs in processes. For example, if a Trojan drops a DLL to disk and then registers a message hook on the system using SefWindowsHookEx, the Trojan can force other processes to load the malicious DLL. Embodiments can identify any newly allocated memory regions in processes after running the malware. This streamlines the procedure of identifying injected code, regardless if it implements sophisticated hiding techniques.

[0216] Table 11 shows the newly loaded DLLs and their respective base addresses, size, and path on disk. Although the VMwareTray.exe process existed both before and after running the suspicious executable, the DLLs were not loaded beforehand. The additions are the result of code that Zeus (the subject malware in this example) injected into all processes on the system. The injected code has dependencies, which explains why several new DLLs exist in VMwareTray.exe after running the suspicious executable.

TABLE 11
+ VMwareTray.exe	0x76bf0000	0xb000	C:\WINDOWS\system32\psapi.dll
+ VMwareTray.exe	0x771b0000	0xa6000	C:\WINDOWS\system32\wininet.dll
+ VMwareTray.exe	0x77a80000	0x94000	C:\WINDOWS\system32\CRYPT32.dll
+ VMwareTray.exe	0x77b20000	0x12000	C:\WINDOWS\system32\MSASN1.dll
+ VMwareTray.exe	0x77120000	0x8c000	C:\WINDOWS\system32\OLEAUT32.dll
+ VMwareTray.exe	0x77ab0000	0x17000	C:\WINDOWS\system32\ws2_32.dll
+ VMwareTray.exe	0x77aa0000	0x8000	C:\WINDOWS\system32\WS2HELP.dll
+ VMwareTray.exe	0x77ad0000	0x9000	C:\WINDOWS\system32\wsock32.dll

[0217] New connects and sockets: Embodiments of the present invention detect newly created connections (to remote hosts) and sockets (listening server ports). It is not possible for rootkits to hide these entries from the malware detection and analysis system, since the evidence comes from a raw memory dump while the infected system is not running. Table 12 shows newly created sockets after running malware Zeus in the provided system. Out of the 4 new sockets on the system, 3 are TCP (type 6) ports 7053, 5972, and 5453. The final new socket is UDP (type 17) port 1051. All sockets are opened by svchost.exe, which indicates that Zeus successfully injected code into svchost.exe to make it open the sockets.

	TABLE 12
	Sockets:
	+ svchost.exe	7053	6	Tue Mar 24 17:20:45 2009
	+ svchost.exe	5972	6	Tue Mar 24 17:20:45 2009
	+ svchost.exe	5453	6	Tue Mar 24 17:20:45 2009
	+ svchost.exe	1051	17	Tue Mar 24 17:20:45 2009

[0218] New file handles: Embodiments may detect if any processes open new files because of the malware executing on the system. This is not the same as detecting newly created files. For example, if a Trojan injects code into explorer.exe and makes explorer.exe call CreateFile( ) on CA\\windows\\system32\\myfile.txt, then the malware detection and analysis output will flag the open handle to myfile.txt in explorer.exe. This works regardless if the Trojan created myfile.txt or if it previously existed.

[0219] Table 13 shows an example of the newly opened file handle output after a Zeus infection. The output clearly indicates that the copy of Zeus operating inside win logon.exe is responsible for “locking” access to the video.dll and audio.dll configuration files, as well as the primary Zeus executable (ntos.exe) by maintaining an open file handle with no sharing.

	TABLE 13
	Files:
	+ winlogon.exe	\WINDOWS\system32\wsnpoem\video.dll
	+ winlogon.exe	\_SYSTEM_64AD0625—
	+ winlogon.exe	\WINDOWS\system32\wsnpoem\audio.dll
	+ winlogon.exe	\WINDOWS\system32\ntos.exe

[0220] New registry keys: Embodiments may detect if any processes open new registry keys because of the malware executing on the system. Similar to the “new files” description above, this is not the same as newly created registry keys. If a process simply opens an existing registry key during the execution of the malware, it shows up in the malware detection and analysis system output.

[0221] New kernel modules: Embodiments may detect if the malware loads any new kernel modules during execution. Embodiments identify kernel drivers that load and then unload quickly, only leaving a thread of execution running in kernel memory. Table 14 shows newly loaded kernel modules. In particular, the new_drv.sys is loaded at 0x00f8a6200 base address and ipnat.sys is loaded at 0xf6dcc000. This malware clearly installs a rootkit, but the malware detection and analysis system easily detects it.

TABLE 14
Modules:
+\??\C:\WINDOWS\new_drv.sys	0x00f8a62000	0x006000	new_drv.sys [ ]
−\SystemRoot\system32\DRIVERS\ipnat.sys	0x00f6dcc00	0x021000	ipnat.sys

[0222] New SSDT entries: The System Service Dispatch Table (SSDT) is a table of function pointers that Windows uses to locate the corresponding kernel functions. The SSDT is the most common place in the kernel for malware to hide API hooks, thus embodiments provide the feature of determining changes in the SSDT after running a suspect binary.

[0223] Table 15 shows newly created or modified SSDT entries. In this example, the malware hooked 3 entries: NtEnumerateValueKey( ) NtQueryDirectoryFile( ) and NtQuerySystemInformation( ). Based on the legitimate purpose of these functions, it indicates that the rootkit attempts to hide registry entries, files on disk, and running processes. Fortunately, the SSDT hooks only hinder code that runs on the active system. As previously described, since embodiments acquire information from the memory dump of an inactive system, the rootkit hooks do not hinder the output.

TABLE 15
SSDT Entries:
0x0049	0x8222858a	NtEnumerateValueKey!UNKNOWN
0x0091	0x822286b6	NtQueryDirectoryFile!UNKNOWN
0x00ad	0x8222885c	NtQuerySystemInformation!UNKNOWN

[0224] The malware detection and analysis system's analysis capabilities via disk access: The information described above comes from parsing the contents of volatile memory (RAM). Embodiments additionally can mount the analysis system's disk as a read-only drive after executing the malware. This allows reliable change detection and extraction capabilities, which are not always possible using just RAM forensics.

[0225] Newly created files: After mounting the analysis system's disk read-only, the malware detection and analysis system can create checksums of all the files. It can easily compute the difference between a baseline disk image to determine which files the malware added, deleted, or modified. The malware detection and analysis system can analyze each newly created file in order to determine its MD5 and SHI hash, file size, file type, contents, and the like. If the file type is an executable, the malware detection and analysis system prints PE header information for the file. If the file type is anything besides an executable, the malware detection and analysis system produces a hexdump of the first 64 bytes of content. Regardless of the file type, the malware detection and analysis system archives all newly created items in a zip file that users can download for additional analysis.

[0226] Tables 16 and 17 show examples of the newly created files output. In Table 16, the computer system dropped a file named ntos.exe. The malware detection and analysis system prints PE header information such as the ImageBase, SizeOfImage, and details on the sections and imported functions. In Table 17, malware Zeus also dropped a binary file named audio.dll. The hexdump makes it easy to see the familiar “LLAH” signature at the beginning of the file, which is the signature of a record header for older malware Zeus configuration files.

TABLE 16
Added: C:\\WINDOWS\system32\ntos.exe
Size: 474112
MD5: d83518b59e544f8322c53c00bc442d9d
SHA1: 860b352fa4ce553d281dc15b40a728b13a857145
Type: MS-DOS executable PE for MS Window's (GUI) Intel 80386 32-bit
TimeDateStamp: 0x4889de1e (Fri Jul 25 14:07:26 2008)
Characteristics: 0x10f
AddressOfEntryPoint: 0x2e4b0
ImageBase: 0x400000
SizeOfImage: 0x30000
DllCharacteristics: 0x0
NumberOfSections: 0x3
Name, VirtualAddress, Misc_VirtualSize, SizeOfRawData,
Characteristics
.bss, 0x1000, 0x22000, 0, 0xe0000080L
.text, 0x23000, 0xc000, 46592, 0xe0000040L
.idata, 0x2f000, 0x1000, 512, 0xc0000040L

TABLE 17
Added: C:\\WINOOWS\system32\wsnpoem\audio.dll
Size: 114
MD5: a3bdbf7a5e20ad6c274d41ff23703c5e
SEA1: 4fe2137558012be158ef18d6af3e2703a5e5f8e3
Type: data
00000000	4c	4c	41	48	6a	00	00	00	fc	09	f2	0d	1a	11	19	27	|LLaHj...........′|
00000010	ec	28	e2	1d	de	fa	e1	28	d6	2b	d2	45	ce	42	ca	49	|.(.....).+.E.B.I.|
00000020	c6	64	c2	66	be	e5	bb	4a	b7	71	bc	4d	ae	53	f7	6c	|.d.f... J.q.M.S.1|
00000030	aa	59	ab	61	e1	9b	f6	98	c9	9e	ca	d2	c6	d5	bf	d7	|.Y.a.............|
00000040

[0227] Newly created registry entries: Embodiments may analyze the raw registry hive files on disk in order to detect changes. After mounting the hard drive for analysis, the malware detection and analysis system copies the hive files and compares the contents with hive files from a baseline image. It reports any deleted entries, modified entries, and added entries. The malware detection and analysis system also creates a zip archive of the raw registry hives that users can download for later analysis. Additionally, the malware detection and analysis system also creates a zip archive of the system event logs (application, system, and security) that users can download.

[0228] Table 18 shows the output of several registry modifications. It indicates that the malware created a new service on the system of type 1 (SERVICE_KERNEL_DRIVER) with a path of C:\WINDOWS\new_drv.sys.

TABLE 18
ADDED	$$$PROTO.HIV\Software\Microsoft\InetData [2009-03-24119:34:32Z]
NEWER	$$$PROTO.HIV\ControlSet001\Service3 [2009-03-24T19:34:31Z]
ADDED	$$$PROTO.HIV\ControlSet001\Services\new_drv [2009-03-24119:34:322]
Type (REG_DWORD) = 0x00000001 (1)
Start (REG_DWORD) =0x00000003 (3)
ErrorControl (REG_DWORD) = 0x00000000 (0)
ImagePath (REG_EXPAND_SZ) = \??\C:\WINDOWS\new_drv.sy3
DisplayName (REG_SZ)=!!!!

[0229] Some embodiments are also capable of acquiring information from other sources besides RAM and the hard drive. For example, embodiments include a usermode API monitor, network traffic analysis modules, the ability to capture the screen throughout an infection, and other functions.

[0230] Usermode API monitor: The API monitor output lists various sources of useful information in studying the exact behavior of an unknown program. It records the following information, without limitation:

[0231] ID of the process making the API call

[0232] Thread ID of the thread making the API call

[0233] Name of the process making the call (i.e. cmd.exe)

[0234] The name of the API call and all relevant parameters and values

[0235] The result of calling GetLastError( ) after the API call

[0236] Table 19 shows an example of the API monitor implemented according to an embodiment of the present invention.

TABLE 19
(0x0), WSASocketW, 0x764, af:0x2;type:0x1 .-protocol:0x0
(0x0), socket, 0x764, af:0x2;type:0x1;protocol:0x0
(0x0), bind, 0x0, socket:0x764
(0xb7), LoadLibraryExW, 0x71ab0000, lpFileName; WS2_32.dll
(0x0), listen, 0x764, socket:0x0
(0x0), CreateThread, 0x778, IpStartAddress:OxSelOf0;lpParameter:0x0;dwCreationFlags:0x0;
(0x0), WaitForSingleObjectEx, 0x0, hHandle:0x77c;dwMilliseconds:0xffffffff
(0x0), inet_addr, 0x100007f, name:127.0.0.1
(0x0), WaitForSingleObjectEx, 0x102, hHandle:0x790;dwMilliseconds:0x0
(0x0), WaitForSingleObjectEx, 0x102, hHandle:0x790;dwMilliseconds:0x0
(0x0), WSASocketW, 0x75c, af:0x2;type:0x1;protocol:0x0
(0x0), socket, 0x75c, af:0x2;type:0x1;protocol:0x0
(0x0), connect, 0x0, socket:0x75c

[0237] Network traffic analysis: Embodiments capture network traffic that enters and exits the analysis system. In some implementations, the system uses a tool for visual breakdown of traffic into protocols, ports, and sessions. For example, users can drill down to any traffic destined for a particular IP address or download individual segments of a packet capture. Table 20 shows the tool's interface implemented by one embodiment.

TABLE 20
TCP/UDP/. . . Sessions
1.	Tue Mar 24 12:20:42 2009	1 s	0.0.0.0:68 <−>	bootps	108	bytes	raw raw1 raw2
			255.255.255.255:67
2.	Tue Mar 24 12:20:42 2009	0 s	172.16.99.254 −>	ICMP	20	bytes	Echo
			172.16.99.130				raw
3.	Tue Mar 24 12:20:42 2009	0 s	172.16.99.254 −>	ICMP	20	bytes	Echo
			172.16.99.130				raw
4.	Tue Mar 24 12:20:43 2009	0 s	172.16.99 130:68 <−>	bootps	108	bytes	raw raw1 raw2
			172.16.99.254:67
5.	Tue Mar 24 12:20:45 2009	8 s	172.16.99.130:1051 <−>	domain	150	bytes	raw raw1 raw2
			172.16.99.2:53				as html
6.	Tue Mar 24 12:20:45 2009	0 s	239.255.255.250:1900 <−>	1053	54	bytes	raw raw1 raw2
			172.16.99.130:1053
7.	Tue Mar 24 12:20:46 2009	6 s	239.255.255.250:1900 <−>	1057	162	bytes	raw raw1 raw2
			172.16.99.130:1057
8.	Tue Mar 24 12:20:46 2009	16 s	172.16.99.255:137 <−>	netbios-ns	1014	bytes	raw raw1 raw2
			172.16.99.130:137
9.	Tue Mar 24 12:20:55 2009	8 s	172.16.99.255:138 <−>	netbios-dgm	432	bytes	raw raw1 raw2
			172.16.99.130:138
10.	Tue Mar 24 12:20:59 2009	0 s	172.16.99.130.1058 <−>	www	0	bytes	raw raw1 raw2
			203.66.154.232:80

[0238] In addition, embodiments may use additional tools to determine if any of the traffic produced by the malware contains already known patterns.

[0239] Custom antivirus signatures: Embodiments may scan uploaded files with multiple anti-virus products. As an example, the malware detection and analysis system can implement YARA (a well-known tool to help identify and classify malware families) and a set of custom-built signatures. The malware detection and analysis system scans any newly created files in addition to any newly allocated memory segments during the forensic RAM analysis. Thus, embodiments use YARA for signature based scanning as well as string-scanning to detect new instances of terms like “bank” and “credit card” in memory after malware executes.

[0240] Table 21 shows how the malware detection and analysis system implements YARA into the analysis in an embodiment. The table shows that malware Zeus injected code into smss.exe, win logon.exe, and services.exe (the others are not shown). In smss.exe, the injected code region starts at 0x00330000 and ends at 0x00351 fff. Based on custom signatures, YARA detected “zbot” (which is another name for malware Zeus) and “inetfwmgr” (which is a generic signature for any code that contains the CLSID for the INetFwMgr interface).

TABLE 21

[0241] Screen captures: Some embodiments are able to capture screen shots of the desktop while malware executes. In one implementation, the system captures a single snapshot if the user requests the feature when submitting a file. In other embodiments, the malware detection and analysis system is capable of taking multiple screen shots across a specified time, which then builds a time-line of thumbnail images. Malware often spawns new processes or creates its own GUI windows during execution that only last a few seconds. Thus, embodiments of the present invention provide the user with the ability to “see” the windows and visual changes in the environment.

[0242] The various functions discussed in relation to Tables 10-21 are not required by all embodiments, but are only provided by way of example. Some embodiments will implement one or more of these various functions. Some embodiments will implement all of the various functions. Therefore, the illustrated functions are not intended to limit embodiments of the present application, but are only provided by way of example to demonstrate the breadth and depth of the analysis made possible by embodiments. One of ordinary skill in the art would recognize many variations, modifications, and alternatives.

[0243] FIG. 10 is an exemplary display screen illustrating malware during operation according to an embodiment. As illustrated in FIG. 10, a screen shot capability is provided by some embodiments. Thus, screen shots of the malware in action are provided. The submitted file opens an application that creates a window on screen for user input.

[0244] FIGS. 12A-12E are exemplary display screens illustrating operation of a system provided according to an embodiment. FIG. 12A illustrates a user interface prior to entry of a submission according to an embodiment. As discussed more fully throughout the present specification, embodiments provide multiple ways to input malicious code. One of the ways is using a command line with a Python (or other suitable) script, which can be used by automated honey pots or other automated malware collections systems. Another way is via email by composing a new message to a computer-attended email address, such as [email protected] or other suitable address, and attaching the malicious code as an attachment to the email.

[0245] Another way illustrated in FIG. 12A is use of a graphical user interface (GUI) that can be associated with a website. When a user makes a submission through the website, the user browses to the file on the user's local file system or other suitable source. The user may also choose a URL to enter rather than uploading a file to the malware detection and analysis system. This method may be used when the user knows the address of a website that is hosting a piece of malware, however, the user does not have a copy of the malware readily available. The GUI provides a user with the ability to enter a command line argument to pass to the piece of malware that is being uploading. For example, if the malware accepts a parameter such as the port on which to listen, the malware may not listen on a port at all if the user does not supply a port (e.g., port 80). Thus, embodiments enable a user to exercise control which command line parameters are sent to the malware. The Email address box allows a user to provide an email address that can receive a report about the malware's behavior after the analysis is complete or other suitable time. The post-execution preference box allows a user to execute a program on the machine after the malware executes. In addition to text entry, embodiments provide a set of preconfigured applications that the user can run after the malware executes, shown in this embodiment in a drop down menu adjacent to the Post execution box. For example, the preconfigured applications may include, Internet Explorer, Firefox, or the like. Additionally, the user may enter the path to any program on the user's system (e.g., Windows) or programs stored remotely from the user's system.

[0246] Providing preconfigured applications like Internet Explorer and Firefox is useful since many malware samples do not exhibit particular behaviors until a browser is opened. For example, this type of malware can wait for the browser to visit a particular website and then the malware will try to steal information from that website. Thus, embodiments provide the ability to visit particular websites with browsers once the malware is installed. The guest operating system can be entered in the configuration screen, for example, XP/SP II, XP/SP III, Vista, Windows 7 other Windows-based operating systems, Macintosh operating systems, mobile operating systems, Linux-based operating systems, and the like. The Guest OS option gives the user the ability to run malware on a different platform to see how it effects those different platforms. Default runtime in the illustrated embodiment is sixty seconds for each piece of malware. Some malware may only execute their malicious behaviors after five minutes, ten minutes, other another period of downtime. Thus, the user is able to increase the runtime to a particular amount of time using the Runtime box in order to capture those behaviors that are only exhibited after a particular amount of time.

[0247] The reboot behavior gives the user the ability to reboot a machine after installing the malware. In some cases, the malware will only trigger particular behaviors after the first reboot. Thus, embodiments give the user the ability to control such behavior. The user can choose to run malware as an administrator or as a normal user. This gives the user the ability to determine how a piece of malware behaves with or without administrative privileges. The user can save a copy of the memory dump using a checkbox or other suitable input device. In some embodiments, the memory dump is saved by default for the memory analysis and the GUI enables the user to download a full copy of the memory dump after operation. The VirusTotal preference allows the user to get a list of a number of antivirus engine signatures for the malware that is uploaded. In an embodiment, a command line program is used to upload the malware sample to VirusTotal, a well-known malware scanning service, and then the system scrapes the webpage from VirusTotal and displays all the results on the malware detection and analysis system webpage.

[0248] In the illustrated embodiment, the debug plug-in is a plug-in for the Win AppDbg Debugger framework although other debugger frameworks are included within the scope of the present invention. Thus, some embodiments provide reverse engineering features as described above. As an example, a scripted debugger can be written and the malware can be analyzed within the context of the script. This gives the user the ability to remove hardware locks from malware. It also gives the user the ability to control each execution stage of the malware with very fine-grained control. Using the Comment box, the user can enter notes or other information related to the malware submission including source of the malware, testing notes, or the like.

[0249] Embodiments provide a selection of priorities for the malware sample uploaded. In the illustrated embodiment of FIGS. 12A-E, a high priority is the default for any submissions made through the web interface. Automated submissions, which are typically received around the clock in larger numbers than manual submissions, can be assigned a lower priority as appropriate to the particular application. After data entry, the user chooses Submit Query to initiate the upload of the malware sample along with the selected and/or default preferences. FIG. 12B illustrates the user interface of FIG. 12A after entry of several preferences according to an embodiment. Other preferences can be entered in other embodiments.

[0250] In an embodiment, once the malware sample has been submitted, the website will automatically redirect the user to a reports information page, which provides the main GUI for reporting on the results of the analysis for the sample that was submitted. Additionally, a GUI can be provided that provides a historical view of all samples that have been submitted to the malware detection and analysis system database. This historical view can include information including, but not limited to: a unique id for each submission; the original file name of the malware; the type of the file that was uploaded; the file status (e.g., waiting, done, defunct, or the like); priority for the malware sample; the run time; API calls monitored with respect to the sample; comments; the date that the file was submitted, or the like. In some embodiments, the historical view color codes the malware submissions a first color during processing and a second color after processing. Additionally, malware can be grouped after analysis using various colors as indicators.

[0251] A search functionality can be provided to enable the user to search for a particular malware sample that has been added in the past. The search can be based on a wide variety of search keys, some of which are listed below. Without limiting embodiments of the invention, the search can be based on:

• • [0252] the malware's MD5 or SHA1 hash
  • [0253] the malware's Ssdeep hash, which is a percent similarity to other variants in the database
  • [0254] the timeframe that the user expects a sample to have been submitted
  • [0255] the user's comments
  • [0256] the type of packer that was used to obfuscate a malware sample
  • [0257] file type (e.g., .exe, .dll, .zip, .sys, svcexe, .svcdll, .pdf, .rar., .chm, .appdll, or the like)
  • [0258] a family name (e.g., particular strings that might exist in an antivirus vendor's name for the malware sample that was uploaded, for example Zeus, Silent Banker, Banker, or the like to find malware with these strings)
  • [0259] registry strings, which allows a user to search for any malware that creates, modifies, or deletes a particular registry key or value based on a substring that the user enters
  • [0260] the file name for any created, modified, or deleted files, including any files loaded into memory rather than just files that are created on the file system
  • [0261] the name of a service (e.g., Windows services) that the malware installs
  • [0262] the name of a host or an IP address contacted by the malware at runtime
  • [0263] any TCP or UDP ports that the malware either contacts a remote host on or that the malware opens on a local machine for incoming connections by other malicious systems
  • [0264] an API function called by the malware sample that was uploaded
  • [0265] an API function hooked by the malware (as discussed in the rootkit detection section above)
  • [0266] the name of a Mutex that was created by malware on the system, which is an effective way of correlating evidence across multiple malware samples. For example, one malware sample could create a particularly named Mutex. The user could then search the database for any other malware samples that created the same named Mutex.

[0267] The system can also provide a list of the behaviors that the malware detection and analysis system detects and the user can search for any samples in the database that exhibit one or more of these behaviors. Reports for the samples analyzed are provided by embodiments on a dashboard provided for each malware sample. In an embodiment, clicking on a file name of a particular malware sample takes the user to the malware sample's home page or dashboard. Information on the malware sample including submission details is provided to the user including, without limitation: the time that it was submitted, the time that the sample was copied to the virtual machine and executed and the time when the analysis was completed, the comment that the user entered during submission, if this sample was a member of any groups and the name of the group if appropriate, the user-entered email address if appropriate, and the like. An option of notifying a new recipient can also be provided.

[0268] Embodiments may provide a screen shot of a virtual machine after executing the malware. If the malware created any type of windows or opened any applications, the user would see them in the screen shot. A severity indicator (a dial, a bar, or the like) can be provided to indicate the severity (e.g., on a scale from 0 to 10) of the piece of malware based on a number of factors, including, without limitation, the number of anti-virus vendors that detected this sample as malicious and the number of behaviors that the malware detection and analysis system detected. Embodiments enable the user to download copies of the reports in one of several formats including a copy of the PDF report. The user can also download a copy of the traffic capture in TCP format for viewing in Wire Shark or any other traffic analysis tool. The user can download a zipped copy of all the files related to the malware sample, including the original file, the screen shot, the PDF report and any files created or dropped by the malware. The user can also view a report for visual breakdown of traffic into protocols, ports, and sessions as illustrated in the discussion of Table 20 above. If the malware sample contacted any hosts during the course of its execution, the user will see a breakout of the TCP, ICMP and UDP sessions. As an example, a malware sample could contact a particular host on port 80 and the user could view the exact requests that this malware sample issued to the remote hosts including binary content represented as text. In an embodiment, the actual binary content is available for download by clicking on an indicator associated with raw entries.

[0269] Embodiments may provide detailed information on the malware sample, which can be referred to as File details. The information can include information collected during the pre-processing stage discussed above. The user interface includes input devices or buttons useful in downloading the original sample (e.g., green arrow button), download an IDA database for the original sample, which allows the user to quickly reverse engineer the file, an information icon. Details include details on the executed file's PE header. In some embodiments, if the system detects any of the values displayed in the information details to be suspicious, based on a pre-populated list of heuristics, then the entries will be highlighted. For example if the file includes resource entries, a Resource Directory field can be highlighted because of the presence of the resource entries in the file. As will be evident to one of skill in the art, this indicates that the executable has other types of content wrapped up inside its own body, which sometimes means that the executable file is going to drop other files when it executes.

[0270] The user can see the TimeDateStamp, which is the compile time of this executable. The user can see all the Section Names and different attributes for them. As an example, for a malware sample, the Section Name “.reloc” can be highlighted as suspicious because its entropy value is very low (e.g., 0.8 on a scale of 0 to 8) indicating the amount of randomness of the data contained within this section. The version information can be extracted from the executable. For one malware sample, the version information includes a claim that the executable was developed by Trend Micro Incorporated and the user can see its internal name. This can be of interest to the user because it shows the user the path on disc that this program existed in before the malware authors distributed it. The malware author's user name can be displayed. Individual resource types that are compiled into the binary of the malware can be displayed. The imported functions can be displayed, which may be a low number for samples that are packed. A list of antivirus vendors that detected this malware as malicious can be displayed as well as the strings extracted from the piece of malware that was uploaded.

[0271] Embodiments of the present invention provide a behavior summary that contains a list of all the behaviors that the malware detection and analysis system detects. The behaviors are highlighted if the malware detection and analysis system detected a particular behavior after running a sample. The behaviors can be based on either the creation of a particular file, the content within the file, some heuristic behavior of the malware, some artifact found in memory after running the malware, or any number of behaviors that the system detects as a result of the malware exhibiting one or more particular behaviors.

[0272] A registry tab can be provided including color coding of the registry entries: For example, green if the entries are new; red if the entries were deleted; orange if the entries were modified; and yellow if the entries are suspicious for any reason. FIG. 12C is an exemplary display screen of a Files tab that can be provided according to an embodiment. The Files tab can use the same color coding for files as the Registries tab. In the illustrated example, the malware sample dropped a file called “user.ds.lll.” The reason this entry is highlighted is because this file is a new file and because the Zeus malware is known to drop files named user.ds.lll. Embodiments may thus use signatures or rules that can indicate suspicious activity based on what has been previously determined in relation to particular piece of malware. In this example, the file size and its type are shown if the file size is not empty. In FIG. 12C, the “sdra.64.exe” file was an MS-DOS executable and the user can see the creation time and the modified time of this file on the NTSF file system. Referring to the entry for the sdra64.exe entry, it is a new file with a timestamp entry that indicates that the malware not only created this file, but once it created it, it tried to hide the fact that it was created by changing the time stamp and making the last modified date in the year 2006. The user can see that the file that was dropped (sdra.64.exe) was detected by antivirus engines and the user is provided with a hex dump preview of the contents of the file that was created. Referring to the entry in FIG. 12C, another file “local.ds” is illustrated. This is the encrypted configuration file that the Zeus malware uses to know which sites it can steal data from. A hex dump is provided, but provides limited information because it is encrypted. Embodiments enable a user to download a copy of the file straight from the website. The system will provide information related to the malware, which can be decrypted on the fly as discussed above. Embodiments provide decryption algorithms that can be used to decrypt a files such as local.ds, which is recognized as an encrypted file from the Zeus malware and decryption can be performed using an encryption key (which can be displayed to the user) and the user can be provided with a display of the plain text information associated with the file. For this local.ds example, the user can view Zeus's configuration file, which a reverse engineer or a customer who was hit by Zeus could use to determine if their institution is targeted. A search functionality can be provided for the decrypted text.

[0273] FIG. 12D is an exemplary display screen of a Timeline tab that can be provided according to an embodiment. The Timeline tab provides information on the timeline for the malware sample. The timeline is composed of artifacts extracted from the file system, the registry, the event log, memory, and the like and is combined into one timeline so that the user can easily see in which order events occurred. Referring to FIG. 12D, the user is presented with the basic timeline and can initially tell that most of the activity was contained within the first few moments of executing the malware. A zoom function is provided so that the user can use the zoom and scroll to the very beginning of the timeline to see the temporal relationship between each event that occurred.

[0274] FIG. 12 E is a zoomed-in portion of the display screen illustrated in FIG. 12D. The original process that was uploaded to the malware detection and analysis system was named 59e7, etc. and the process was assigned process ID 1988. That happened at point A. At point B, this particular file was created, which you can be illustrated by a full path to the file. At point C another process was started with the exact same name as the original process, but with a different process ID. This indicates that the Zeus malware created a copy of itself running in memory, most likely to evade debuggers that are trying to capture all the activity created by this process, so it spawned another copy of itself to be able to execute outside the control of a debugger. At point D, the creation of the sdra.64.exe file occurred and at point E local.ds was created. Thus, the timeline provides insight into events and their correlation, including the creation of files, events, registry entries and the like.

[0275] Embodiments may provide an Event Log tab that contains extracted entries from the application, system, and security event logs from the virtual machine on which the malware ran. The user can see the event IDs, along with the exact time that the event occurred and the message that was logged into the event log. If the user clicks a hyperlink provided by the system, the user is taken to a website that displays more information about the message that was logged. In one example, the user can see a message indicating that svchost.exe was used to create a listening port on the system, IPv4 protocol and TCP protocol. Thus, as a result of running the malware, svchost.exe started listening on a port on the system, which is evidence that the Zeus malware may open up a backdoor on the system. The system can provide, through a Network tab, a listing of host names that were resolved via DNS or that the malware executed. The user can see a hex dump of the traffic that went across the wire, the destination and source ports, the size of the payload, and the like.

[0276] In an embodiment, all of the host names with which the malware communicates are converted to longitude and latitude using geographical IP libraries and then they are converted to a static image along with the name of the city in which the destination IP existed. Thus, the user is provided with a quick view of the countries or continents that the malware contacts during its command and control sessions.

[0277] Embodiments may provide a grouping feature. This feature allows a user to upload a zip file that contains multiple specimens. Each file will be analyzed individually or in conjunction in a virtual machine and then those files will be grouped together so that later on the user can tell that all the grouped files were part of the same incident. In an embodiment, historical files that have been added and are part of a group are indicated with a green circle that stands for the parent of a group, that is, the initial file that was uploaded that later resulted in dropped or downloaded files that were analyzed individually. The user can click on a family tree button and be provided with a view of the original sample that was uploaded to the malware detection and analysis system. The unique ID assigned to the sample can be displayed The family tree can indicate that when the sample executed, it performed a variety of actions including creation of other files, creation of sockets, starting of processes, creation of registry keys, and the like. All the artifacts that occurred on the system as a result of running the malware sample can be shown by the family tree. Analysis of files that were dropped in response to the execution of the parent file is performed and similar information is provided at lower levels of the family tree in association with the dropped files.

[0278] Embodiments may provide views of family trees based on behaviors rather than artifacts, providing the user with insight into the behaviors discusses above. In the behavior view, clicking on the name of an will enable the user to view the full report for that file.

[0279] Another tab that may be provided by embodiments is the Memory tab. The Memory tab provides the user with a listing of processes that have been created, for example, in a highlighted format. New processes can be highlighted as of interest. Additionally, programs such as YARA can be used to indicate that a new process is malicious, for example, because antivirus signatures are detected in the memory of this process. Information related to the process can be provided to the user, for example, the offset within the process memory where the antivirus signature was detected, the reason why these particular strings in the process memory are of interest, and the like. An unpacked copy of the process can be downloaded or the user can click an information button to see the visible strings in the unpacked process memory.

[0280] For some malicious files, a process may be executed and then terminated before the list of processes is compared. However, embodiments of the present invention can detect the injection of code into other processes, which is how the malware stays persistent on the machine. As an example, code could be injected into svchost.exe, which is detected as suspicious strings in the memory of svchost.exe. As a result of the running the malware, svchost.exe can open a handle to a particular dll on the machine, which is not in the InLoad order, InMemory order, or InInit order module lists, indicating that these dlls have been loaded into svchost.exe but they are being hidden by a rootkit. Additionally, embodiments of the present invention provide a list of the API hooks in kernel modules that the rootkit installs. For example, NtEnumerateKey API hides registry values and NtSaveKey API and NtSaveKeyEx API prevent other programs on the system from overriding the values that the rootkit uses to stay persistent. The user can see the kernel modules that were loaded as a result of running the rootkit.

[0281] Embodiments may enable a user to see a list of the sockets that malware opens, for example, within the memory of svchost.exe, providing the user with the list of listening ports that were created as a result of running the malware. Additionally, embodiments may enable a user to view the processes that malware can hook, for example, the wuauclt.exe process using an IAT hook. Embodiments also can show the user EAT and inline hooks. The user can see the function name and the containing module and where code should exist for NT query directory file and the location that it's being redirected to. When the user clicks on a hooked function, the user can see a hex dump of the data that exists at that address and can see a disassembly of the instructions that execute instead of the real function. Thus, in contrast with conventional methods, embodiments of the present invention provide the user with a list of API functions are hooked, as well as a hex dump and a disassembly of the rootkit instructions.

[0282] Embodiments may provide summaries of system performance, including the most common behaviors, a list of the most common packers, the network command and control distribution in the form of a map, an antivirus word cloud, and the like.

[0283] Other embodiments of the invention will be apparent to those skilled in the art from consideration of the specification and practice of the embodiments disclosed herein, including, without limitation, the ability to embody the various embodiments in instructions on a computer-readable medium, which when executed perform a method comprising a malware detection and analysis system. In particular, it should be appreciated that the processes defined herein are merely exemplary, and that the steps of the processes need not necessarily be performed in the order presented. It is intended that the specification and examples be considered as exemplary only, with a true scope and spirit of the embodiments being indicated by the following claims.