Patent Grant
10382448
The present invention relates generally to the field of computer system security. More specifically, providing a system and method for detecting and mitigating command injection attacks from occurring on a network, system, device, or media.
Command injection attacks exploit flaws in software. The present inventors recognize that one solution to prevent such attacks would be to avoid software flaws by using safe programming practices or programming constructs that do not allow such flaws. While this approach may be technically feasible, in many instances it is not practical. First, it would be impractical to redesign or re-implement the large body of legacy software that already exists. Second, even for newly-developed software, time-to-market pressure favors the quick delivery of new features over careful security considerations. Third, many software applications are produced by programmers that have not been properly trained in best security practices. And fourth, a software application is often created by composition with other software components whose provenance and adherence to security best practices can be of dubious quality. In short, not only are command injection attacks severe, they are here to stay for the foreseeable future.
An aspect of an embodiment of the present invention provides, among other things, a method, system, and computer readable medium for protecting software against a class of attacks known as “command injection attacks” that exploit common vulnerabilities found in software. The #1 and #2 listed items on MITRE's 2011 list of the “Top 25 Most Dangerous Software Errors” (https://cwe.mitre.org/top25/index.html) and the #1 and #2 listed flaws on the Open Web Application Security Project “Top 10 Application Security Risks” (https://www.owasp.org/index.php/Top_10_2010-Main) enable attackers to craft command injection attacks. Similarly, the #4 listed item on MITRE's list-“Improper Neutralization of Input During Web Page Generation (‘Cross-site Scripting’)”—enables attacker to craft command injection attacks. Still yet, the #23 listed item on MITRE's list-“Uncontrolled Format String”—enables attacker to craft command injection attacks.
These attacks are dangerous because they will frequently allow attackers to completely take over the software, steal data, or prevent the software from working at all.
Accordingly, an aspect of an invention of the present invention provides an efficient solution for thwarting command injection attacks and for allowing software to continue operating despite attempts at subverting software.
An aspect of the present invention provides for, but not limited thereto, an approach for preventing or mitigating command injection attacks; and accordingly it recognizes that various attacks inject a payload that then gets processed by the software to carry out the attackers' intentions. The problem is that software vulnerabilities allow the malicious attacker payload to be processed normally as if it were a non-malicious payload.
The attack payload can be viewed as foreign genetic material that will typically not be present in the native software. Thus, one approach of stopping command injection attacks is to track all external data processed by software, e.g., network input, data entered in a web form, files uploaded to a server, and monitor its use in security-critical commands. If such data is detected, the software can refuse to carry out the command. Such an approach is referred to as taint propagation. The present inventors recognize that while such an approach may be effective in terms of stopping command injection attacks, this approach is not practical as tracking the flow of data through a software program is expensive and severely impacts performance.
Instead of keeping track of external (potentially malicious) data, an aspect of an embodiment of the present invention provides a method, system, and computer readable medium to identify native genetic material, such as but not limited thereto, command fragments. An aspect of an embodiment of the present invention scans software for fragments that can be used to form commands issued by the software. Whenever the software issues a command, the method or system of the present invention can intercept the command and uses a matching algorithm or method to reassemble the command using the identified fragments. The method or system of the present invention detects an attempted attack by looking for critical parts of the command that do not originate from the previously identified fragments. Once an attack is detected, the method or system of the present invention performs an analysis to determine the type of attacks, and then is able to remediate the attack to allow the software to continue normal operation.
By identifying native genetic material used to form a command (instead of attempting to track foreign material), an aspect of an embodiment of the present invention provides an approach that dispenses with expensive taint propagation techniques.
Furthermore, an aspect of an embodiment of the present invention provides an approach that is transparent and can be automatically applied to software already installed on a computing system, without necessarily involving the original software manufacturer, and without requiring modifications to the software program to be protected.
Accordingly, an aspect of various embodiments of the present invention is that it would have wide applicability. In particular, an aspect of an embodiment of the present invention can be used to secure any software that is network-enabled. An aspect of an embodiment of the present invention protects software against a severe class of attacks known as command injection attacks. Examples of such attacks include Structured Query Language (SQL) Injection Attacks (e.g., database attack), operating system (OS) Command Injection Attacks, Cross-Site Scripting Attacks, LDAP Injection Attacks, XML Injection Attacks, Cross-Site Scripting (XSS) Attacks, and format string attacks. Command injection attacks exploit the #1, #2, #4 and #23 software errors as described in MITRE's list of Top 25 Most Dangerous Software Errors. Further, an aspect of an embodiment of the present invention also allows many software programs to continue normal operation even after an attempted attack.
An aspect of an embodiment of the present invention protects software even after the software manufacturer has shipped it. An aspect of an embodiment of the present invention does not require users to modify the program. An aspect of an embodiment of the present invention can even be applied directly to programs shipped in binary form. Furthermore, modern software makes frequent use of third-party components whose quality may be unknown or untrusted. Accordingly, an aspect of an embodiment of the present invention allows such software to be built with security protection despite these third party components.
The novel aspects of the present invention provides, but not limited thereto, a process and algorithms (and related system and computer readable medium) for determining fragments, reassembling fragments to match commands to determine whether a command is safe or suspicious, such as a potential attack, and allowing software to continue execution despite potential attacks or actual attacks. This may all be done transparently, efficiently, without requiring any efforts from software developers.
An aspect of various embodiments of the present invention may provide a number of advantages, such as but not limited thereto, the ability to protect software without requiring users to modify the software, which therefore makes it a practical solution.
An aspect of an embodiment of the present invention thwarts a severe and important class of attacks, namely command injection attacks. These attacks exploit software vulnerabilities that are in the MITRE's list of the “Top 25 Most Dangerous Software Errors”. An aspect of an embodiment of the present invention also allows many software programs to continue normal operation even after an attempted attack.
An aspect of various embodiments of the present invention may be utilized for a number of products and services, such as but not limited thereto, as discussed below. An aspect of an embodiment of the present invention is applicable to all software. It can be applied on a wide variety of platforms and applications, including laptops, desktops, cell phones, tablets, photo frames, cameras, video recorders, PDAs, routers, web browsers and servers, music players, televisions, game consoles, handheld gaming devices, critical infrastructure devices (traffic lights, power relays), heart-rate monitors, biometric monitors, networked-enabled sensors, etc. The importance of the invention will only rise over time as more and more devices are becoming network-enabled, e.g. refrigerator, home thermostats, automobiles, medical devices, etc. Further, an aspect of an embodiment can be applied on a wide variety of applications for infrastructure related systems and devices, such as power utilities, water utilities, public utilities, office infrastructure, home dwellings, department of defense infrastructure and systems, military equipment and systems, medical equipment and systems, vehicles, aircraft, server farms, and watercraft.
An aspect of an embodiment of the present invention provides a method, system and computer readable medium that can thwart OS command injection attacks (as well as data base or SQL injection attacks, database and web based application; format string attacks; LDAP injection attacks, XPATH injection attacks, cross-site scripting (XSS) attacks, attacks against scripting languages, and NoSQL injection attacks or the like) by matching the program's signature fragments to the commands it attempts to issue. If commands cannot be matched, the software, system and method of an embodiment of the present invention assumes that the command that has been injected into the program is potentially dangerous.
An aspect of an embodiment of the present invention provides, among other things, a method, system and computer readable medium that provides, among other things, taint inference and positive tainting resulting in positive taint inference
An aspect of an embodiment of the present invention is that it does not require the software manufacturer to change their code. In fact, as discussed above, an aspect of an embodiment of the invention can be applied even after the manufacturer has shipped the software.
An aspect of an embodiment of the present invention provides, but not limited thereto, a computer method for detecting command injection attacks. The method may comprise: receiving software code; extracting string fragments from the received software code to provide extracted signature fragments; receiving command instructions; converting the received command instructions into command fragments; identifying critical parts from the commands fragments; determining if the critical parts are untrusted or trusted by matching with the extracted signature fragments; identifying potential attacks upon the condition that a command includes critical parts that are untrusted; and communicating the identification of potential attacks to an output device.
An aspect of an embodiment of the present invention provides, but not limited thereto, a computer method for detecting command injection attacks. The method may comprise: receiving software code; receiving string fragments to provide signature fragments; receiving command instructions; converting the received command instructions into command fragments; identifying critical parts from the commands fragments; determining untrusted or trusted parts of the command instructions by using the signature fragments; identifying potential attacks upon the condition that a command includes critical parts that are untrusted; and communicating the identification of potential attacks to an output device.
An aspect of an embodiment of the present invention provides, but not limited thereto, a system for detecting command injection attacks based on command instructions to be received from a client processor or client data memory (or other processor or data memory as desired, needed or required). The system may comprise: a memory unit operative to store software code; and a processor. The processor may be configured to: extract string fragments from the software code to provide extracted signature fragments; receive the client command instructions; convert the received command instructions into command fragments; identify critical parts from the commands fragments; determine if the critical parts are untrusted or trusted by matching with the extracted signature fragments; identify potential attacks upon the condition that a command includes critical parts that are untrusted; and communicate the identification of potential attacks to an output device.
An aspect of an embodiment of the present invention provides, but not limited thereto, a system for detecting command injection attacks based on command instructions to be received from a client processor or client data memory (or other processor or data memory as desired, needed or required). The system may comprise: a memory unit operative to store software code and a processor. The processor may be configured: receive string fragments to provide signature fragments; receive command instructions; convert the received command instructions into command fragments; identify critical parts from the commands fragments; determine untrusted or trusted parts of the command instructions by using the signature fragments; identify potential attacks upon the condition that a command includes critical parts that are untrusted; and communicate the identification of potential attacks to an output device.
An aspect of an embodiment of the present invention provides, but not limited thereto, a non-transitory computer readable medium including instructions executable by a processor for detecting command injection attacks. The instructions may comprise: receiving software code; extracting string fragments from the received software code to provide extracted signature fragments; receiving command instructions; converting the received command instructions into command fragments; identifying critical parts from the commands fragments; determining if the critical parts are untrusted or trusted by matching with the extracted signature fragments; identifying potential attacks upon the condition that a command includes critical parts that are untrusted; and communicating the identification of potential attacks to an output device.
An aspect of an embodiment of the present invention provides, but not limited thereto, a non-transitory computer readable medium including instructions executable by a processor for detecting command injection attacks. The instructions may comprise: receiving software code; receiving string fragments to provide signature fragments; receiving command instructions; converting the received command instructions into command fragments; identifying critical parts from the commands fragments; determining untrusted or trusted parts of the command instructions by using the signature fragments; identifying potential attacks upon the condition that a command includes critical parts that are untrusted; and communicating the identification of potential attacks to an output device.
An aspect of an embodiment of the present invention provides, but not limited thereto, methods and systems that are described herein for detecting command injection attacks. A positive, taint inference method includes receiving signature fragments on one hand, converting command injection instructions into command fragments on another hand, thus identifying potential attacks upon the condition that a command injection instruction includes critical untrusted parts by using signature fragments. A system detects command injection attacks using this kind of method, and remediates and rejects potential attacks.
These and other objects, along with advantages and features of various aspects of embodiments of the invention disclosed herein, will be made more apparent from the description, drawings and claims that follow.
The accompanying drawings, which are incorporated into and form a part of the instant specification, illustrate several aspects and embodiments of the present invention and, together with the description herein, serve to explain the principles of the invention. The drawings are provided only for the purpose of illustrating select embodiments of the invention and are not to be construed as limiting the invention.
Still referring to
Additionally, a command parsing module 314 is provided for identifying the security-critical parts of a command 316. Some examples of critical parts of a command 316 may include, for example, critical tokens and keywords. Accordingly, this shall parse the intercepted security-critical commands 312 to identify the security-critical parts of a command 316, such as critical tokens and keywords. For OS commands, the security critical parts 316 may consist of command names, options, delimiters, special characters and the setting of environment variables. This command parsing module 314 is responsible for identifying the security-critical parts of a command. For example, for OS commands may include implementing a simple, combined lexical analyzer and parser.
Additionally, a trust inference module 318 is provided for determining the parts of the command that are untrusted or trusted. In an approach, the trust inference module 318 makes this determination by matching or comparing the intercepted commands 312 against the extracted signature fragments 306. If a match is found then that part of the command of the signature fragment is considered trusted according to the trust inference module 318. Conceptually, the trust inference component 318 infers which portions of the command come from within the program, and which ones come from external sources (i.e., potentially malicious). In another embodiment, the trust inference component may use different policies for establishing trust. Examples include requiring that critical parts of a command come from one signature, or from a set of related signatures, or using any available annotations.
Accordingly, still referring to
For instance, where a potential attack is detected 324, then the command may be rejected 334 with an error message 335 or remediated 336 for instances where remediation is deemed appropriate. The remediation may include altering the command 338 so as to deem it acceptable and therefore acceptable for transmission to the command interpreter 332 (e.g., a server, such as a backend server).
A command response 340 is transmitted from the command interpreter 332 (for example, operating system) to the runtime software 310.
Still referring to
Still referring to
Next, referring to
Still referring to
Various embodiments or aspects of the invention may be implemented as software in a computing device, or alternatively, on hardware. For example,
In addition to implementation on a standalone computing machine, embodiments of the invention may also be implemented on a network system comprising a plurality of computing devices that are in communication with a networking means, such as a network with an infrastructure or an ad hoc network. The network connection may be wired, wireless, or a combination thereof.
As a way of example,
Main memory 134 also may be used for storing temporary variables or other intermediate information during execution of instructions to be executed by processor 138. Computer system 140 further includes a Read Only Memory (ROM) 136 (or other non-volatile memory), or other static storage device coupled to a bus 137 for storing static information and instructions for processor 138. A storage device 135 may be coupled to the bus 137 for storing information and instructions. The storage device 135 may include a magnetic disk or optical disk, a hard disk drive for reading from and writing to a hard disk, a magnetic disk drive for reading from and writing to a magnetic disk, and/or an optical disk drive (such as DVD) for reading from and writing to a removable optical disk. The hard disk drive, magnetic disk drive, and optical disk drive may be connected to the system bus by a hard disk drive interface, a magnetic disk drive interface, and an optical disk drive interface, respectively. The drives and their associated computer-readable media provide non-volatile storage of computer readable instructions, data structures, program modules and other data for the general purpose computing devices. Typically computer system 140 includes an Operating System (OS) stored in a non-volatile storage for managing the computer resources and provides the applications and programs with an access to the computer resources and interfaces. An operating system commonly processes system data and user input, and responds by allocating and managing tasks and internal system resources, such as controlling and allocating memory, prioritizing system requests, controlling input and output devices, facilitating networking and managing files. Non-limiting examples of operating systems are Microsoft Windows, Mac OS X, and Linux.
The term “processor” is meant to include any integrated circuit or other electronic device (or collection of devices) capable of performing an operation on at least one instruction including, without limitation, Reduced Instruction Set Core (RISC) processors, CISC microprocessors, Microcontroller Units (MCUs), CISC-based Central Processing Units (CPUs), and Digital Signal Processors (DSPs). The hardware of such devices may be integrated onto a single substrate (e.g., silicon “die”), or distributed among two or more substrates. Furthermore, various functional aspects of the processor may be implemented solely as software or firmware associated with the processor.
Computer system 140 may be coupled via bus 137 to a display 131, such as a Cathode Ray Tube (CRT), a Liquid Crystal Display (LCD), a flat screen monitor, a touch screen monitor or similar means for displaying text and graphical data to a user. The display may be connected via a video adapter for supporting the display. The display allows a user to view, enter, and/or edit information that is relevant to the operation of the system. An input device 132, including alphanumeric and other keys, may be coupled to bus 137 for communicating information and command selections to processor 138. Another type of user input device is cursor control 133, such as a mouse, a trackball, or cursor direction keys for communicating direction information and command selections to processor 138 and for controlling cursor movement on display 131. This input device typically has two degrees of freedom in two axes, a first axis (e.g., x) and a second axis (e.g., y), that allows the device to specify positions in a plane.
The computer system 140 may be used for implementing the methods and techniques described herein. According to one embodiment, those methods and techniques are performed by computer system 140 in response to processor 138 executing one or more sequences of one or more instructions contained in main memory 134. Such instructions may be read into main memory 134 from another computer-readable medium, such as storage device 135. Execution of the sequences of instructions contained in main memory 134 causes processor 138 to perform the process steps described herein. In alternative embodiments, hard-wired circuitry may be used in place of or in combination with software instructions to implement the arrangement. Thus, embodiments of the invention are not limited to any specific combination of hardware circuitry and software.
The term “computer-readable medium” (or “machine-readable medium”) as used herein is an extensible term that refers to any medium or any memory, that participates in providing instructions to a processor, (such as processor 138) for execution, or any mechanism for storing or transmitting information in a form readable by a machine (e.g., a computer). Such a medium may store computer-executable instructions to be executed by a processing element and/or control logic, and data which is manipulated by a processing element and/or control logic, and may take many forms, including but not limited to, non-volatile medium, volatile medium, and transmission medium. Transmission media includes coaxial cables, copper wire and fiber optics, including the wires that comprise bus 137. Transmission media can also take the form of acoustic or light waves, such as those generated during radio wave and infrared data communications, or other form of propagated signals (e.g., carrier waves, infrared signals, digital signals, etc.). Common forms of computer-readable media include, for example, a floppy disk, a flexible disk, hard disk, magnetic tape, or any other magnetic medium, a CD-ROM, any other optical medium, punch-cards, paper-tape, any other physical medium with patterns of holes, a RAM, a PROM, and EPROM, a FLASH-EPROM, any other memory chip or cartridge, a carrier wave as described hereinafter, or any other medium from which a computer can read.
Various forms of computer-readable media may be involved in carrying one or more sequences of one or more instructions to processor 138 for execution. For example, the instructions may initially be carried on a magnetic disk of a remote computer. The remote computer can load the instructions into its dynamic memory and send the instructions over a telephone line using a modem. A modem local to computer system 140 can receive the data on the telephone line and use an infra-red transmitter to convert the data to an infra-red signal. An infra-red detector may receive the data carried in the infra-red signal and appropriate circuitry can place the data on bus 137. Bus 137 carries the data to main memory 134, from which processor 138 retrieves and executes the instructions. The instructions received by main memory 134 may optionally be stored on storage device 135 either before or after execution by processor 138.
Computer system 140 also may include a communication interface 141 coupled to bus 137. Communication interface 141 provides a two-way data communication coupling to a network link 139 that is connected to a local network 111. For example, communication interface 141 may be an Integrated Services Digital Network (ISDN) card or a modem to provide a data communication connection to a corresponding type of telephone line. As another non-limiting example, communication interface 141 may be a local area network (LAN) card to provide a data communication connection to a compatible LAN. For example, Ethernet based connection based on IEEE 802.3 standard may be used such as 10/100BaseT, 1000BaseT (gigabit Ethernet), 10 gigabit Ethernet (10 GE or 10 GbE or 10 GigE per IEEE Std 802.3ae-2002 as standard), 40 Gigabit Ethernet (40 GbE), or 100 Gigabit Ethernet (100 GbE as per Ethernet standard IEEE P802.3ba), as described in Cisco Systems, Inc. Publication number 1-587005-001-3 (6/99), “Internetworking Technologies Handbook”, Chapter 7: “Ethernet Technologies”, pages 7-1 to 7-38, which is incorporated in its entirety for all purposes as if fully set forth herein. In such a case, the communication interface 141 typically include a LAN transceiver or a modem, such as Standard Microsystems Corporation (SMSC) LAN91C111 10/100 Ethernet transceiver described in the Standard Microsystems Corporation (SMSC) data-sheet “LAN91C111 10/100 Non-PCI Ethernet Single Chip MAC+PHY” Data-Sheet, Rev. 15 (02-20-04), which is incorporated in its entirety for all purposes as if fully set forth herein.
Wireless links may also be implemented. In any such implementation, communication interface 141 sends and receives electrical, electromagnetic or optical signals that carry digital data streams representing various types of information.
Network link 139 typically provides data communication through one or more networks to other data devices. For example, network link 139 may provide a connection through local network 111 to a host computer or to data equipment operated by an Internet Service Provider (ISP) 142. ISP 142 in turn provides data communication services through the world wide packet data communication network Internet 11. Local network 111 and Internet 11 both use electrical, electromagnetic or optical signals that carry digital data streams. The signals through the various networks and the signals on the network link 139 and through the communication interface 141, which carry the digital data to and from computer system 140, are exemplary forms of carrier waves transporting the information.
The processor 138 may execute received code as it is received, and/or stored in storage device 135, or other non-volatile storage for later execution. In this manner, computer system 140 may obtain application code in the form of a carrier wave.
The concept of instruction location randomization may be implemented and utilized with the related processors, networks, computer systems, internet, modules, and components and functions according to the schemes disclosed herein.
Referring to
Referring to
Referring to
It should be appreciated that an embodiment may entail a system that comprises an assembly of hardware component modules (or firmware) configured to perform the functions of the enumerated steps.
Practice of an aspect of an embodiment (or embodiments) of the invention will be still more fully understood from the following examples and experimental results, which are presented herein for illustration only and should not be construed as limiting the invention in any way.
A. Architecture and Related Method
The command interception component or module 208 intercepts security-critical commands so that they can be vetted. The trust inference component or module 218 determines which characters in the intercepted command should be trusted by matching the command against the extracted string fragments, i.e., signature fragment 206. Any unmatched character is deemed untrusted. Combining fragments native to the protected binary to infer trust is a novel form of positive taint inference and one of the advantages associated with an embodiment the present invention architecture.
The command parsing component or module 214 parses the intercepted command to identify critical tokens and keywords. The attack detection component or module 220 combines the output of the trust inference component/module 218 and command parsing component/module 214 to determine whether an attack has occurred. A command is deemed an attack if a critical token or keyword is marked as untrusted.
Upon attack detection, an embodiment of the present invention system or method that either rejects 234 the command outright and returns an error code 235, or it alters 238 the command before passing it on to the operating system 232 (or other command interpreter). Altering may entail changing a part of the command or removing a part of the command. The current prototype uses a simple form of error virtualization that simulates a failed command invocation by substituting an error code in place of the actual command [See 24], [See 25].
To illustrate how an embodiment may work, the present inventors use the following vulnerable program as a working example:
When the program in the working example is passed a benign input such as “README”, the resulting command is shown in the first line of
In this case, /bin/cat is trusted as it matches the composition of the signature fragments “/bin” and “/cat” extracted in the offline signature Fragment Extraction process (see
C. Example with Attack Input
Consider the malicious input README; rm -fr * that seeks to recursively delete user files. The resulting command is shown in
Software, System and Method Detailed Overview
While the present illustration architecture is generic, e.g., it could be applied to web applications, mobile applications, or the like as discussed throughout the present disclosure, the present inventors present details and discuss challenges encountered as the inventors map the software, system and method into a practical instantiation to defeat OS command injection attacks for binary programs.
A. Signature Fragment Extraction
The accuracy of the fragment extraction process may be crucial. If fragments are missed, valid commands might be flagged as injections. If extra fragments are extracted, malicious command injections might not be flagged (See Example Set No. 4 at Section B for further discussion).
1) String Extraction:
Extracting string fragments from binary programs is more difficult than it first appears. The present inventors' first attempt used the Linux program strings, which linearly scans a binary program and extracts null-terminated sequences of ASCII characters that have a length larger than a given threshold. Unfortunately, short strings are sometimes important. Consider this C++ snippet:
string q=“rm”;
q+−“-f”;
q+−filename;
system(q.c_str ( ));
which creates and executes an OS command.
Using strings, the threshold needs to be sufficiently low to find short strings. Unfortunately, low thresholds tend to yield lots of garbage strings, which affects accuracy. Furthermore, compilers use many optimizations that can make strings harder to detect. For example, to initialize a string on the stack, a compiler might use a sequence of store instructions:
mov [esp+28], 0x2d206d72 # “rm-”
mov [esp+32], 0x00002066 # “f \0\0”
Each move stores four bytes onto the stack, ultimately creating the proper null-terminated string. Other compiler idioms may complicate accurately finding all strings, as well. The present inventors have seen examples of the compiler inlining some standard library functions that have constant operands, such as memcpy (dst, “rm -f”, 6). This optimization yields in-lined constants much like the previous string initialization example. Lastly, strings reports all strings in the executable file, which can include debug information, shared library names, compiler-version identifiers, etc. As these types of strings cannot be used to form OS commands, they should be excluded from consideration.
To deal with these issues, the present inventors use static analysis of the program to derive the string fragments. The static analysis starts by fully disassembling the program into a database which holds each instruction in the program, indexable by address, function, and control flow information. The present inventors use a hybrid linear-scan disassembler and recursive-descent disassembler to ensure we get good coverage of all instructions, as described by Hiser. et al. [See 26], [See 27].
After disassembly is complete, the instructions are scanned for accesses or creation of string values. The present inventors analyze each instruction's immediate operands and apply three heuristics to identify string fragments:
Finally, we check for other string fragments or string pointers in data sections.
2) Post-Processing of Signature Fragments:
Programs compiled from C or C++ often contain statements that use format specifiers, e.g. % d, % f, % s, % x. We split such fragments into their constituent sub-fragments using the format specifiers as delimiters. A fragment such as
“/bin/rm -f % s; /bin/touch % s”
would be split into the sub-fragments “/bin/rm -f”, and “; /bin/t ouch”. As the analysis cannot be sure that such fragments are used as format strings, the original fragment as well as the sub-fragments are retained in the list of signatures.
The present inventors noticed the short fragments that contain potentially dangerous shell metacharacters (fragments 306-315), or short fragments that could be composed in an attack (fragments 273-288). In Example Set No. 2 at Section E and Example Set No. 3, the present inventors discuss how the software, system and method policies deal with short and potentially dangerous fragments.
B. Command Interception
For binaries derived from C/C++, commands are typically encapsulated in an Application Programming Interface (API) and accessed via dynamically-linked shared libraries.
The software, method and system prototype may leverages standard library interposition facilities to transparently intercept and wrap function calls to the underlying operating system. The software, method and system intercepts the system, popen, rcmd, and exec family of functions. Other functions could be intercepted as well, but the present inventors have identified these as the primary candidates for OS command injection.
C. Command Parsing
This component is responsible for identifying the security-critical parts of a command. For OS commands, the critical parts consist of command names, options, delimiters, and the setting of environment variables.
The software, method and system prototype uses a simple, combined lexical analyzer and parser. The parser is careful to identify special characters which could indicate the start of a new command (such as the semicolon character), match quotation marks and parentheses, etc. Ideally, the present inventors would use a full, formally-verified lexical analyzer and distinct parser to detect keywords, etc. However, it may be difficult due to the nature of the shell language (bash in our case). Consider this command:
echo Touching $(file); touch ‘foobar’
What are the “correct” lexical analysis and parse for this command? The answer depends on the value of the file variable and the output of the foobar executable. If file is set to a single quote character and foobar returns the same thing, then there is exactly one command, echo. Since variables are expanded and sub-processes are executed before the command is parsed, the correct parse cannot be determined a priori. Under most circumstances, though, such odd substitutions are not the case.
For the purposes of detecting OS command injections, we need to know the possible places where a command could be invoked. The present system or method may provide a simple parser for the subject prototype assumes that the structure of the command is not changed by the results of executing subcommands. In the case of the present inventors' simple example, the parser marks the command like so:
where C indicates that a critical command character exists at the given location.
D. Trust Inference
Conceptually, the Trust Inference component infers which portions of the command come from within the program, and which ones come from external sources. To accomplish this step, it checks each substring in the command to determine if that location is within the set of signature fragments. This pseudocode illustrates the process:
for each signature fragment, f
Conceptually, this algorithm and related method could be quite expensive, O (n3) where n=max(length(sig), #sigs, length(command)). In practice, though, the present inventors use a move-to-front heuristic to organize the signature fragments required to trust commands and exit the outermost loop when enough of the command is trusted to verify its safety. Further, each command and each signature fragment is typically short, on the orders of tens or hundreds of characters. These simple observations and adjustments dramatically reduce the time necessary to make the inference. Example Set No. 3 at Section D empirically evaluates the overhead associated with inferring trust markings.
E. Attack Detection
Attack detection may consist of scanning the command for any character that has been marked as untrusted by the Trust Inference component and critical by the Command Parsing component. In addition, it may desirable to impose the constraints shown in
This policy helps to compensate for the case when a short, critical token, such as a semi-colon or a quotation mark, is present in the set of signature fragments. Such fragments allow attackers great latitude to create strings that append new commands, as in “; rm -rf”. Unfortunately, these signature fragments cannot simply be discarded, because many programs do use such fragments to terminate their commands. However, it appears uncommon for a program to use such fragments to introduce a new command, so the present inventors disallow this behavior entirely.
When no attack is detected then the software, method and system passes the command to the operating system to execute. However, if an attack is detected, the software, method and system do not pass through the command, but can enact any one of a variety of remediation responses, such as shutting down the program, warning the user and asking for permission to continue, or logging the attack. For the prototype described in this example set, the present inventors chose to return an error code as if the library call had failed. This policy makes sense in many cases, as well-written programs are designed to gracefully handle error conditions.
Example Set No. 4 discusses potential sources of false negatives and false positives, and their implications in further detail.
Evaluation
To evaluate the security and performance of an aspect of an embodiment of the present invention system and method the present inventors have prototyped a system and applied it to a variety of engineered and real-world benchmarks. This Example Set describes the Experimental Setup, Benchmarks, Performance Evaluation and Security Evaluation in more detail.
A. Experimental Setup
For an evaluation, the present inventors used a 32-bit Virtual Box virtual machine running Ubuntu 12.04 with 4 GB of RAM and a 2 GHz Xeon E5-2620 processor.
B. Benchmarks
To evaluate the performance and security of an aspect of an embodiment of the present invention system and method, the present inventors have collected a variety of benchmarks. For real-world benchmarks with CVE reports, the present inventors used the SpamAssassin Milter Plugin [See 28], an email filler interface for detecting spam, and cbrPager [See 29] version 0.9.16. a program to decompress and view high-resolution images. The present inventors configured SpamAssassin Milter version 0.3.1 with SpamAssassin version 3.3.2 and Postfix version 2.9.6. Both of these programs have real-world OS command injection vulnerabilities.
The present inventors also used a set of vulnerable programs independently developed by MITRE Corporation from real-world, open-source software. Each program was seeded with a command injection vulnerability. This process was repeated to create many variants with the vulnerability at many locations. Each variant has inputs that represent normal program input, as well as exploit inputs.
Finally, the present inventors used a set of small exploitable programs, most less than 100 lines, that were developed by Raytheon. Like the MITRE programs, normal and exploit inputs are provided.
Lastly, to help evaluate performance, the present inventors developed a series of microbenchmarks. These benchmarks create an OS command from command line input, and use a tight loop to execute that command as frequently as possible, doing no other work. There are two dimensions of variation in the micro benchmarks: 1) the command to be executed and 2) the primitive used to invoke the command. There are two possible commands to be executed. The command to be executed in one case is echo hello, and in the second case is bzip2 dickens.txt [See 30], [See 31]. The two cases represent a fast command and a somewhat more reasonable workload that compresses a 775 KB file. Each microbenchmark uses one of the following primitives to invoke the command: execv, popen, or system.
C. Security Evaluation
The present inventors used a combination of programs with reported real-world vulnerabilities, synthetic test programs, and real-world programs seeded with vulnerabilities by an independent testing team to evaluate the strength of an aspect of an embodiment of the present invention system and method.
1) Real-World Attacks: the present inventors evaluated an aspect of an embodiment of the present invention system and method against two reported command-injection vulnerabilities that the present inventors were able to reproduce in open-source binaries.
The first attack, based on CVE-2008-2575, is a command injection in cbrPager [See 32]. To extract images, cbrPager invokes the system library call to execute unzip or unrar on the archive, without sanitizing the filename. By crafting an input such as “; rm -rf *;”.cbr and providing it where a filename is expected, cbrPager is tricked into executing a malicious command when it attempts to open the putative file. An aspect of an embodiment of the present invention system and method is to have the ability to detect attempts to open a malicious filename and return an error from the system library call. These actions result in the program displaying a message that the file cannot be opened, and exiting harmlessly.
The second attack, based on CVE-2010-1 132, is a remote exploit in the SpamAssassin Milter Plugin [See 4] (spamass-milter), which integrates the SpamAssassin spam filter with either sendmail or Postfix. The vulnerability occurs when the milter is invoked with the -x “expand” option, to pass the email address through alias and virtusertable expansion to allow emails to be redirected to other accounts. In this case, the popen function is invoked on sendmail with the email address provided from SMTP as an argument, without properly sanitizing the email address, which can contain a pipe character. With an SMTP command such as RCP T TO: <username+: “|rm /var/spool/mail”>, arbitrary commands can be executed; with careful crafting, these may be sufficient to open a remote shell. This particular embodiment of the system and method was able to harmlessly block any command injections. The signatures extracted from spamass-milter do not include the vertical bar (pipe) character, foiling any attempt to exploit this weakness. Moreover, the Milter plugin properly error-checks the popen function call, so it continues to function without loss of service in the face of an attempted exploit.
2) Synthetic Attacks: The present inventors evaluated an aspect of an embodiment of the present invention system and method against engineered test suites developed by Raytheon and independently by MITRE. The Raytheon engineered suite consists of 18 microtests demonstrating command injections with 22 good inputs and 35 bad inputs, using 9 different function calls ([f] exec [1, 1e, 1p, v, ve, vp], system and popen) and a variety of input-processing techniques. This particular embodiment of the system and method mitigates all of the bad inputs while breaking none of the good inputs in this test suite.
The MITRE test suite includes 477 OS command injection (based on CWE-78) and 516 OS argument injection (based on CWE-88) test cases [See 33]. These test cases are based on inserting vulnerabilities into seven base programs: Cherokee, grep, nginx, tepdump, wget, w3c (from libwww), and zsh. Each test case involves inserting a vulnerable call to popen at various locations in the base program. For the CWE-78 test cases, this call invokes nslookup with an unsanitized argument specified from an environment variable or untrusted file. For the CWE-88, the program builds the command using the format string “find/-iname % s.” Semicolon characters are properly sanitized when constructing the command, but the user can still include input that has a -exec argument that is ultimately passed to find. Consequently, they could use an input such as “*-exec rm { }\;” to remove files or execute other commands. For each test case, ten good inputs and two bad inputs are provided. In each case, an aspect of an embodiment of the present invention system and method was able to intercept the bad inputs without altering behavior on the good inputs.
D. Performance Evaluation
The present inventors selected two of the benchmarks from the MITRE suite where the seeded vulnerability was in the main loop of a server; most vulnerabilities were injected into startup or shutdown code, and there was no significant performance difference. The seeded vulnerability was set to execute only once, but for timing purposes the present inventors modified the code slightly so that it executed on every request to the server. The benchmarks are based on Cherokee (C-C078-CHER-04-DF09-02) and nginx (C-C078-NGIN-04-DT03-02), two production-quality web servers. An aspect of an embodiment of the present invention system and method performed 50 timings, each consisting of downloading a small html file (574 bytes). Even with the seeded vulnerability in the main loop and the small download size, no statistically significant difference in timing was observed with this particular embodiment of the system and method.
For SpamAssassin Milter, the present inventors wrote a simple client that uses gettimeofday to measure the time spent in processing an email transaction. The present inventors also modified cbrPager to measure the time to render the first page of a 49 MB input file. Like the MITRE benchmarks, these benchmarks show no statistically significant overhead.
Unfortunately, the server applications have relatively high variance due to network latencies, disk caching, etc. To deal with this issue and benchmark worst-case overhead, the present inventors use the microbenchmarks described in Example Set No. 3 at Section C2. For these benchmarks, an aspect of an embodiment of the present invention system and method performs 50 timings, where each timing invokes 10 or 1,000 OS commands for the bzip2 or echo microbenchmark, respectively. The microbenchmarks that invoke bzip2 to compress a file show that an aspect of an embodiment of the present invention system and method causes practically no overhead, only 0.2%. The true worst-case performance overhead is when the program does nothing but issue OS commands, and each OS command invocation completes extremely quickly. This case is represented by the microbenchmark that issues the echo hello command. These benchmarks show that the absolute worst case overhead might be as high as 22%. However, in practice the actual work performed by the program and by executing the OS command clearly dominates the overall run-time. Only our worst-case microbenchmarks demonstrate that a particular embodiment of the system and method generates any measurable overhead.
To verify the move-to-front heuristic was working properly, the present inventors measured the overhead of the echo microbenchmark that uses the system function to invoke OS commands as the present inventors vary the number of signature fragments. The present inventors automatically added randomly generated strings to the program's DNA fragments.
In practice, this additional overhead is negligible since most programs have few signatures. For example, SpamAssassin Milter has 316 signatures and nginx has 2,017.
Based on these microbenchmark and real-world benchmark performance results, the present inventors believe that in practice an aspect of an embodiment of the present invention system and method would introduce no measurable overhead, and is the fastest OS command injection detector to date.
E. Analysis Time
An aspect of an embodiment of the present invention system and method measured the time for offline analysis (i.e., signature Fragment Extraction) of the real-world benchmarks. The results are shown in the table of
In summary, an aspect of an embodiment of the present invention system and method provides, among other things, a new, efficient, approach for detecting faint markings based on positive taint inference. The present inventors' findings indicate that an aspect of the particular system and method can be effectively used to detect OS command injection attacks on binary programs. Furthermore, an aspect of an embodiment of the present invention system and method has demonstrated that it can be used in many real-world situations because it has negligible performance overhead and can be applied directly to binary programs without need for source code or compiler support.
4.1 Signature Fragment Extraction
The goal of signature fragment extraction is provided to extract string fragments, i.e., signature fragments, from the binary software and its associated libraries. It may include steps of Signature Fragment Mining and Signature Analysis and Filtering.
(a) Signature Fragment Mining
A purpose of this step is to analyze the software and extract possible command fragments from the software program. In a subsequent step, the signature fragment matching step, command fragments will be reassembled using a variety of techniques to match parts of commands issued by the software program.
Depending on how the software is delivered, e.g. as a binary file, a binary file with a set of libraries, as a set of source files, as byte code, as assembly language code, as a text file in a scripting language, different methods may be used to identify potential fragments. These methods include:
(b) Signature Fragment Analysis and Filtering
A purpose of this step is to further process signature fragments to aid in the Signature Matching process (later described below in this Set Number in the signature fragment matching section). This is an example of post-processing. This step may include:
Note that this step is optional as signature fragments generated in the Signature Fragment Mining step may be used directly.
4.2 Trust Inference (AKA Signature Fragment Matching)
A purpose of this step is to analyze commands emitted by the software program and locate signature fragments that are contained within the command. For example, a program may emit the following command to access a database:
command: SELECT * from users where userid=‘john’;
If the fragments included the following:
fragment 1: bob the frog
fragment 2: SELECT * from users
fragment 3: where userid=′
fragment 4: ′;
Then the matching process would find that the following parts of the command were made up from fragments (shown underlined):
SELECT * from users where userid=‘john’;
In this example, the only part of the command that is not matched by a string fragment would be: john
Another convention of representing the same effect would be as follows:
Then the matching process would find that the following parts of the command were made up from fragments (shown with ‘T’s for trusted, ‘U’s for untrusted):
SELECT * from users where userid=‘john’;
TTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTUUUUTT
There are many possible ways to find fragments within a command. Examples include:
For the remainder of this exposition, the present inventors will refer to the portions of the command that have been matched against a signature fragment as vetted. As mentioned above, another convention would be designated as trusted (as designated with a ‘T’).
The signature fragment matching process may be carried out in several ways, including:
This approach may be implemented with a variety of techniques, including using source-to-source transformers, static rewriters, dynamic rewriters.
4.3 Detection
A purpose of this step is to detect suspicious commands that may indicate that an entity is carrying out an attack or is otherwise trying to manipulate emission of various commands emitted by the software program.
Detection is performed by analyzing a command and checking whether critical parts of the command have been vetted in the signature fragment matching step. If a critical part of a command has not been vetted, then this provides a strong indication that an attack is underway.
(a) Command Parsing
A command parsing component is provided for identifying the security-critical parts of a command. Some examples of critical parts of a command may include, for example, critical tokens and keywords.
Determining what is critical depends on the nature of the command and what resources may be accessed via the command. Consider the case of queries to a database expressed in the Structured Query Language (SQL). The present inventors use two different commands.
Command 1: select * from users where userid=‘john’;
Command 2: SELECT * from users where userid=′ ‘OR 1=1; -- john’;
Then the command parsing process would find that the following parts of the command were security-critical parts (shown with ‘C’ for critical; and left blank if not critical):
Note that in many cases establishing whether a given portion of a command is deemed critical (and should be checked to see whether it is vetted) may involve a process known in the computer science literature as parsing a string with respect to a given language. This process is somewhat analogous to identifying the parts of speech in a language such as English. Using this analogy, one could deem verbs (but not nouns) to be critical. In the first example, the string john is not deemed critical as it corresponds to data once the command is parsed with respect to SQL. However, the string OR is deemed critical as it corresponds to an instruction in the query. In fact, this query would return all users contained in the database, which would represent a severe leakage of information.
(b) Attack Detection
One could use the policy that SQL keywords or tokens contained in the command should be vetted.
In the first example (vetted portions of the command are shown underlined),
SELECT * from users where userid=′john′;
all relevant SQL keywords tokens are vetted, therefore the command is deemed valid. In the second example,
SELECT * from users where userid=′ ′ OR 1 = 1; -- john′;
there are multiple parts of the commands that are not vetted (i.e., not matched and therefore designated as untrusted, U) and that correspond to a critical SQL keyword or token (i.e., critical ant therefore designated a critical, C): ′ OR=; --
This command is therefore deemed suspicious.
If a command is deemed normal (nothing suspicious has been detected), then the query can be issued normally. However, if a command is deemed suspicious, several remediation policies may be applied (Example Set No. 4 at Section 4.5).
4.4 Validation
Note that the detection step may yield both false positives and false negatives.
For example, if the signature fragment matching step does not identify a critical part of a command as vetted, then the detection step will deem the command suspicious when it should not. The error represents a false positive.
If the signature fragment mining or the signature fragment analysis and filtering steps yield a signature fragment that would be used in an attack, then this signature fragment would be vetted, and therefore the malicious command would evade detection. This error represents a false negative.
Several techniques may be used to minimize false positives, including:
These patterns may then be used to generate further signature fragments;
Several techniques may be used to minimize false negatives, including:
Note that when seeking to minimize false positives or negatives, or in general, when using signature fragments, any of the information associated with the fragments may be used. Finally, the validation step is optional, as the rate of false negatives or positives may already fall within acceptable ranges.
4.5 Remediation
Once a command is deemed suspicious, a wide array of policies may be applied.
Example policies include:
Once a command is deemed suspicious, or at a desired time or predetermined time, a wide array of policies may be applied.
Example policies include:
The monitoring may include methods like “expensive dynamic analysis.” It may be more efficient to do it on suspicious command injections, but it is not a requirement.
In this embodiment of the present invention, the software program is written in the C language. While this embodiment of the invention operates directly on the compiled binary form of the program, the C source code is shown for a simple example program to aid in the understanding of Applicant's invention. In this example, the program issues commands and accesses results from a back-end Postgres database server via several functions: PQfinish, PQconnectdb, PQexec, PQstatus, PQresultStatus. PQClear. The set of functions used to access the database server is often referred to as an API, or Application Programming Interface. While the present inventors show examples of API functions for the Postgres database, the present inventors note that other databases typically offer their own API functions.
In general, an API will be available for any commands of interest.
5.1 C Source Code
#include <stdio.h>
#include <stdlib.h>
#include “libpq-fe.h”
void exitNicely(PGconn *p_conn)
{
PQfinish (p_conn);
exit (1);
}
int main(int argc, char **argv)
{
char conninfo[1024];
char query[1024];
PGconn *conn;
PGresult *res;
sprintf(conninfo, “dbname=% s”, getenv(“PGDATABASE”));
conn=PQconnectdb(conninfo);
if (PQstatus(conn) !=CONNECTION_OK)
{
}
// this line is vulnerable to SQL injection
sprintf(query, “select * from doip where comment=‘% s’;”, argv[1]);
res=PQexec(conn, query);
if (PQresultStatus(res)==PGRES_TUPLES_OK)
{
}
else
{
}
PQclear(res);
PQfinish(conn);
return 0;
}
5.2 Signature Fragment Extraction
To identify the initial set of signature fragments an embodiment of the present invention method and system may process the binary that results from compiling the C source code example with the Unix strings utility. Possible fragments produced are:
[^_]
_bss_start
Connection to database failed: % s
dbname=% s
_edata
_end
exit
_fini
fprintf
getenv
GLIBC_2.0
GLIBC_2.4
_gmon_start—
_init
_IO_stdin_used
_Jv_RegisterClasses
libc.so.6
_libc_start_main
/lib/ld-linux.so.2
libpq.so.5
PGDATABASE
PQclear
PQconnectdb
PQerrorMessage
PQexec
PQfinish
PQresultStatus
PQstatus
PTRh0
Query failed: % s
Query success: % s
select * from doip where comment=‘% s’;
sprintf
_stack_chk_fail
stderr
Note that many of the strings in the C source code are successfully extracted from the program binary.
(b) Signature Fragment Analysis and Filtering
In this step of post processing, signature fragments that correspond to function names are filtered out, along with common symbol names and strings that correspond to well-known libraries. This can be done using the nm facility in Unix or other tools to extract symbol name information. Furthermore, signature fragments that contain % s or other format string specifiers can be replaced with signature fragments that are generated by removing the % s. For example, the fragment:
select * from doip where comment=‘% s’;
can be replaced with the following fragments:
select * from doip where comment=′
′;
The final set of signature fragments produced by this step is:
select * from doip where comment=′
Connection to database failed:
Query success:
Query failed:
PGDATABASE
dbname=
PTRh0
[^_]
′;
The signature fragments may then be stored in a file (or other form of persistent storage), or in memory, so that they can be used during the signature fragment matching step. Alternatively, signature fragments may denote a pattern to be used in the matching process.
5.3 Trust Inference (AKA Signature Fragment Matching)
Using the interposition facilities in modern Unix systems, API function calls related to issuing commands to the database are intercepted. As an example of interposition, the code below illustrates how calls to PQexec, one of the major function for issuing commands to the database, are intercepted:
PGresult* PQexec(PGconn* p_conn, const char *p_query)
{
static PGresult* (*my_pgExec) (PGconn*, const char *)=NULL;
if (!my_pgExec)
{
}
char *errMsg=NULL;
if (sqlfw_verify(p_query, &errMsg))
{
}
else
{
}
}
This code intercepts the calls to PQexec which then enables the signature fragment matching, the detection and the remediation steps. The function sqlfw_verify ( ) encodes the signature fragment matching and detection steps. If the query command (p_query) is deemed normal then the query is passed through to the original function that implements PQexec. Otherwise, if the query command is deemed to be suspicious, then a remediation policy is applied.
Consider the case where the program input yields a command query that is not an attack:
select * from doip where comment=‘bob’;
The fragment matching process would mark the following characters as vetted (shown underlined, and can also be designated as trusted, ‘T’, or untrusted, ‘U’.):
select * from doip where comment=‘bob’;
TTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTUUUTT
In the case when the command query is an attack, e.g.:
select * from doip where comment=′ ‘or 1=1; -- bob’;
The attack would result in a leak of all entries from the doip table. The fragment matching process would mark the following characters as vetted (shown underlined, and can also be designated as trusted, ‘T’ or untrusted, ‘U’.):
select * from doip where comment=′ ‘or 1=1; --bob’;
TTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTUUUUUUUUUUUUUUUTT
5.4 Detection
To determine whether the query command is malicious, Applicant parses the query string to look for critical parts (shown with ‘C’ for critical; and left blank if not critical) of the commands that are not vetted. In the case of the normal query:
select * from doip where comment = ‘bob’;
All critical parts of the command, i.e., SQL keywords and tokens such as select from where =‘ ’; are all vetted.
However, in the case of an attack:
select * from doip where comment = ‘’ or 1=1; --bob’;
there are multiple SQL keywords and tokens, i.e., ′ or =; -- that are not vetted.
If a command is deemed normal, i.e. non-malicious, the present method and system may go ahead and let the command be issued. Otherwise, if a command is deemed suspicious, i.e. a potential attack, the user or system may perform some remediation action.
5.5 Remediation
Once an attack is detected, there are multiple remediation actions that are possible. In this example, the present inventors generate a malformed query command intently, execute this malformed query, and return any resulting errors to the software program.
5.6 Validation
In this example, the present method and system may does not attempt to perform any validation of the detection algorithm. However, if a regression test were supplied with the software program, the present method and system may execute the program with known non-malicious test inputs. If any inputs resulted in a detection of an attack, then the inputs could be analyzed to generate further signature fragments. Alternatively, the present inventors could create tests manually or by recording all inputs given to the program during a user session.
If attack inputs were supplied or published, the present inventors could verify that the detection algorithm detected the attack. If the detection step did not detect the attack, we could judiciously remove signature fragments until the attack was detected.
In this embodiment, the original software program is also written in the C language. However, as was the case with the embodiment discussed in Example Set No. 3, an embodiment of the present invention may operate directly on the compiled binary form of the program. In the example shown below, the program issues operating system commands using API functions contained in the standard C libraries, e.g. system.
6.1 C Source Code
#include <stdio.h>
int main(int argc, char *argv[ ])
{
if (argc <2)
{
}
char command[2048];
sprintf(command, “/bin/ls % s”, argv[1]);
int ret=system(command);
fprintf (stdout, “% s returned with code: % d\n”, command, ret);
return 0;
}
6.2 Signature Fragment Extraction
(a) Signature Fragment Mining
To identify the initial set of signature fragments an embodiment of the present method and system may process the binary that results from compiling the C source code example with the Unix strings utility. The fragments produced are:
[^_]
/bin/ls % s
fprintf
fwrite
GLIBC_2.0
GLIBC_2.4
_gmon_start—
_IO_stdin_used
libc.so.6
_libc_start_main
/lib/ld-linux.so.2
must specify at least one argument
PTRh
QVh4
sprintf
% s returned with code: % d
_stack_chk_fail
stderr
stdout
system
(b) Signature Fragment Analysis and Filtering (e.g., Post-Processing).
Using similar techniques as those described in Section (b) of this Example Set No. 6, the final set of signature fragments produced is:
[^_]
/bin/ls
must specify at least one argument
PTRh
QVh4
returned with code:
6.3 Trust Inference (AKA Signature Fragment Matching)
Using the interposition facilities in modern Unix systems, API function calls related to issuing operating systems commands are intercepted. These functions include:
system, popen, rcmd, fexecve, execve, execl, execle, execv, execvp, execvpe, execlp.
Consider the case when the command is not an attack, for example when a file is supplied to the program. The resulting command would be:
/bin/ls myFile
The fragment matching process would mark the following characters as vetted (shown underlined, and can also be designated as trusted, ‘T’ or untrusted, ‘U’.):
/bin/ls myFile
TTTTTTT UUUUUU
In the case when the command query is an attack, e.g., the program is supplied with the input someOtherFile; cat /etc/passwd
The resulting command would be:
/bin/ls someOtherFile; cat /etc/passwd
and would potentially leak the content of the password file contained on the user's machine. The fragment matching process would mark the following characters as vetted (shown underlined, and can also be designated as trusted, ‘T’ or untrusted ‘U’.):
/bin/ls someOtherFile; cat/etc/passwd
TTTTTTTTUUUUUUUUUUUUUUUUUUUUUUUUUUUUUU
6.4 Detection
To determine whether the command is malicious, the present method and system may look for critical parts (critical character designated with “C” and non-critical designated by leaving blank) of the commands that are not vetted. In the example:
/bin/ls myFile
TTTTTTTTUUUUUU
CCCCCCC
The critical parts (designated with ‘C’) of the command /bin/ls is marked as vetted (i.e. underlined or with the alternative convention, trusted), and therefore this command is deemed non-malicious.
However, in the case of the following command,
/bin/ls someOtherFile; cat /etc/passwd
The shell program cat follows the; terminator symbol and is deemed critical. Since cat is not vetted (i.e. without underline or with the alternative convention, untrusted), the command is deemed suspicious.
An embodiment of the current prototype treats the first shell program in the command as critical, and well as any shell programs that follow a separator symbol such as ;, |, ∥, &&.
In addition, shell program options and the name of environment parameters are deemed critical. Thus in the following command, the characters shown in boldface are deemed critical:
MYENV=someValue /bin/ls someFile && rm -fr /
6.5 Validation
In this embodiment, validation is not performed. If desired validation would proceed as described in Example Set No. 4 at Section 4.5 and Example Set No. 5 at Section 5.5.
The present inventors note that while the SQL injection and OS injection attack examples are presented separately, they can, in practice, be combined into one protection mechanism or approach. In general, an aspect of an embodiment of the present invention allows for the interception of arbitrary API functions, and thus it is in fact quite easy to compose various protections against a wide variety of command injection attacks. Further examples of command injection attacks include: LDAP Injection Attacks, XPATH Injection Attacks, Cross-Site (XSS) Scripting Attacks, Log Injection Attacks, and Format String Attacks. In general, aspects of various embodiments of the present invention may apply to any injection attacks wherein an attacker can manipulate critical parts of a command.
The examples provided illustrated the use of an embodiment of the invention for program binaries compiled from C source code. As was already described in Example Set No. 4, the present inventors note that an aspect of an embodiment of the present invention is generally applicable to, but not limited thereto, a variety of software program types, including software expressed in various scripting languages (Ruby, Python, PHP, Perl, etc.), byte code (e.g. Java byte code, Microsoft's Common Intermediary Language, Dalvik byte code), assembly language, etc. An aspect of an embodiment of the present invention also applies to software program binaries generated from a wide variety of languages, including C and C++, and other languages that are compiled to binary form.
An aspect of an embodiment of the present invention provides, but not limited thereto, a computer method for detecting command injection attacks. The method may comprise: receiving software code; extracting string fragments from the received software code to provide extracted signature fragments; receiving command instructions; converting the received command instructions into command fragments; identifying critical parts from the commands fragments; determining if the critical parts are untrusted or trusted by matching with the extracted signature fragments; identifying potential attacks upon the condition that a command includes critical parts that are untrusted; and communicating the identification of potential attacks to an output device.
The method of example 1, further comprises: remediating or rejecting one or more of said identified potential attack commands.
The method of example 2, wherein said remediation includes altering the identified potential attack command.
The method of example 2 (as well as subject matter of example 3), further comprising: transmitting said remediated command to a command interpreter module.
The method of example 4 (as well as subject matter of one or more of any combination of examples 2-3), further comprising: executing said remediated command in said command interpreter; and generating a command response.
The method of example 2 (as well as subject matter of one or more of any combination of examples 3-5), further comprising: transmitting said remediated command for execution; and generating said executed command as a command response.
The method of example 2 (as well as subject matter of one or more of any combination of examples 3-6), wherein said remediation of said identified potential attack command includes providing one or more of the following instructions: terminating the software program, repairing the potential attack command, enabling additional system monitoring, enabling additional analysis, logging that a suspicious command is being attempted, alerting a supervisory remote computer that a breach is being attempted, notifying an administrator of the suspicious command, not issuing the command but returning an error code, not issuing the command but returning a non-error code, issuing a known bad command and returning the resulting error codes, displaying an alert to the end user, or any combination thereof.
The method of example 1 (as well as subject matter of one or more of any combination of examples 2-7), wherein said identifying critical parts includes at least one of the following: parsing, dissecting, lexical analyzing, or tokenizing.
The method of example 1 (as well as subject matter of one or more of any combination of examples 2-8), wherein said extracting comprises a string extracting technique, and said method further comprises: post-processing of said extracted signature fragments.
The method of example 9 (as well as subject matter of one or more of any combination of examples 2-8), wherein said post-processing may comprise one or more of the following: removing said extracted signature fragments; adding said extracted signature fragments; modifying said extracted signature fragments; or annotating said extracted signature fragments with additional information
The method of example 1 (as well as subject matter of one or more of any combination of examples 2-10), wherein said identifying potential attacks includes using annotations.
The method of example 11 (as well as subject matter of one or more of any combination of examples 2-10), wherein said annotations are generated by said post-processing step.
The method of example 1 (as well as subject matter of one or more of any combination of examples 2-12), wherein said signature fragments specify patterns.
The method of example 13 (as well as subject matter of one or more of any combination of examples 2-12), wherein said patterns comprise one or more of the following: regular expressions, wild-card specifiers, format string specifiers, context-free grammars, or grammars.
The method of example 1 (as well as subject matter of one or more of any combination of examples 2-14), wherein if the condition to identify that potential attacks have occurred has not been satisfied then said command is deemed as safe, thereby defining a safe command.
The method of example 15 (as well as subject matter of one or more of any combination of examples 2-14), further comprising: transmitting said safe command to a command interpreter module.
The method of example 16 (as well as subject matter of one or more of any combination of examples 2-15), further comprising: executing said safe command; and generating a command response.
The method of example 15 (as well as subject matter of one or more of any combination of examples 2-14 and 16-17), further comprising accepting said safe command.
The method of example 1 (as well as subject matter of one or more of any combination of examples 2-18), wherein said output device includes at least one of the following: storage, memory, network, printer or a display.
The method of example 1 (as well as subject matter of one or more of any combination of examples 2-19), wherein said command instructions comprise instructions to an operating system (OS).
The method of example 1 (as well as subject matter of one or more of any combination of examples 2-10), wherein said command instructions comprise database commands or structured query language (SQL) instructions.
The method of example 1 (as well as subject matter of one or more of any combination of examples 2-21), wherein said command instructions comprise instructions to a format string interpreter.
The method of example 1 (as well as subject matter of one or more of any combination of examples 2-22), wherein said command instructions comprise instruction to an LDAP interpreter.
The method of example 1 (as well as subject matter of one or more of any combination of examples 2-23), wherein said command instructions comprise instruction to an XPATH interpreter.
The method of example 1 (as well as subject matter of one or more of any combination of examples 2-24), wherein said command instructions comprise instructions to a web language.
The method of example 1 (as well as subject matter of one or more of any combination of examples 2-25), wherein said command instructions comprise instructions to a scripting language interpreter.
The method of example 1 (as well as subject matter of one or more of any combination of examples 2-26), wherein said command instructions comprise instructions to a NoSQL database.
The method of example 1 (as well as subject matter of one or more of any combination of examples 2-28), wherein said command instructions include any combination of one or more of the following: instructions to an operating system (OS), instructions to a data base or instructions to an, SQL interpreter, instructions to a web based language; instructions to a format string interpreter; instructions to a LDAP, interpreter, instructions to an XPath interpreter, instructions to a scripting language, and instructions to a NoSQL database.
An aspect of an embodiment of the present invention provides, but not limited thereto, a computer method for detecting command injection attacks. The method may comprise: receiving software code; receiving string fragments to provide signature fragments; receiving command instructions; converting the received command instructions into command fragments; identifying critical parts from the commands fragments; determining untrusted or trusted parts of the command instructions by using the signature fragments; identifying potential attacks upon the condition that a command includes critical parts that are untrusted; and communicating the identification of potential attacks to an output device.
The method of example 29, wherein said receiving string fragments further comprises extracting string fragments from said received software.
The method of example 29 (as well as subject matter of example 30), wherein said receiving string fragments includes one or more of any of the following: receiving from the software developer, downloading from a networked resource, distributed with the software, manually specifying string fragments, or static or dynamic analysis of the software code.
The method of example 29 (as well as subject matter of one or more of any combination of examples 30-31), further comprises: monitoring, remediating or rejecting one or more of said identified potential attack commands.
The method of example 32 (as well as subject matter of one or more of any combination of examples 30-31), wherein said remediation includes altering the identified potential attack command.
The method of example 32 (as well as subject matter of one or more of any combination of examples 30, 31 and 33), further comprising: transmitting said remediated command to a command interpreter module.
The method of example 34 (as well as subject matter of one or more of any combination of examples 31-33), further comprising: executing said remediated command in said command interpreter; and generating a command response.
The method of example 32 (as well as subject matter of one or more of any combination of examples 30, 31 and 33-35), further comprising: transmitting said remediated command for execution; and generating said executed command as a command response.
The method of example 32 (as well as subject matter of one or more of any combination of examples 30, 31, and 33-36), wherein said monitoring, remediation or rejection of said identified potential attack command includes performing one or more of the following actions: terminating the software program, repairing the potential attack command, enabling additional system monitoring, enabling additional analysis, logging that a suspicious command is being attempted, alerting a supervisory remote computer that a breach is being attempted, notifying an administrator of the suspicious command, not issuing the command but returning an error code, not issuing the command but returning a non-error code, issuing a known bad command and returning the resulting error codes, displaying an alert to the end user, or any combination thereof.
The method of example 29 (as well as subject matter of one or more of any combination of examples 30-37), wherein said identifying critical parts includes at least one of the following: parsing, dissecting, scanning, tokenizing, or lexical analysis.
The method of example 29 (as well as subject matter of one or more of any combination of examples 30-38), further comprising post-processing of said received string fragments.
The method of example 39 (as well as subject matter of one or more of any combination of examples 30-38), wherein said post-processing may comprise one or more of the following: removing some of said string fragments; adding additional fragments to said string fragments; modifying said string fragments; or annotating said string fragments with additional information.
The method of example 29 (as well as subject matter of one or more of any combination of examples 30-40), wherein said determining untrusted or trusted parts of the command instructions includes using a trust-verification policy.
The method of example 41 (as well as subject matter of one or more of any combination of examples 30-40), wherein said trust-verification policy includes ensuring that certain critical parts of said command instructions exactly match one signature fragment.
The method of example 29 (as well as subject matter of one or more of any combination of examples 30-42), wherein said identifying potential attacks includes using a trust-verification policy.
The method of example 43 (as well as subject matter of one or more of any combination of examples 30-42), wherein said trust-verification policy includes ensuring that certain critical parts of said command instructions exactly match one signature fragment.
The method of example 40 (as well as subject matter of one or more of any combination of examples 30-39 and 41-44), wherein said identifying potential attacks includes using annotations.
The method of example 29 (as well as subject matter of one or more of any combination of examples 30-45), wherein said identifying potential attacks includes using annotations.
The method of example 45 (as well as subject matter of one or more of any combination of examples 30-44 and 46), wherein said annotations are generated by said post-processing step.
The method of example 29 (as well as subject matter of one or more of any combination of examples 30-47), wherein said signature fragments specify patterns.
The method of example 48 (as well as subject matter of one or more of any combination of examples 30-47), wherein said patterns comprise one or more of the following: regular expressions, wild-card specifiers, format string specifiers, context-free grammars, or grammars.
The method of example 29 (as well as subject matter of one or more of any combination of examples 30-49), wherein if the method of identifying potential attacks results in no potential attacks then said command is deemed as safe, thereby defining a safe command.
The method of example 50 (as well as subject matter of one or more of any combination of examples 30-49), further comprising: transmitting said safe command to a command interpreter module.
The method of example 51 (as well as subject matter of one or more of any combination of examples 30-50), further comprising: executing said safe command; and generating a command response.
The method of example 50 (as well as subject matter of one or more of any combination of examples 30-49 and 51-52), further comprising accepting said safe command.
The method of example 29 (as well as subject matter of one or more of any combination of examples 30-53), wherein said output device includes at least one of the following: storage, memory, network, printer or a display.
The method of example 29 (as well as subject matter of one or more of any combination of examples 30-54), wherein said command instructions comprise instructions to an operating system (OS).
The method of example 29 (as well as subject matter of one or more of any combination of examples 30-55), wherein said command instructions comprise database commands or structured query language (SQL) instructions.
The method of example 29 (as well as subject matter of one or more of any combination of examples 30-56), wherein said command instructions comprise instructions to a format string interpreter.
The method of example 29 (as well as subject matter of one or more of any combination of examples 30-57), wherein said command instructions comprise instruction to an LDAP interpreter.
The method of example 29, wherein said command instructions comprise instruction to an XPATH interpreter.
The method of example 29 (as well as subject matter of one or more of any combination of examples 30-59), wherein said command instructions comprise instructions to a web language.
The method of example 29 (as well as subject matter of one or more of any combination of examples 30-60), wherein said command instructions comprise instructions to a scripting language interpreter.
The method of example 29 (as well as subject matter of one or more of any combination of examples 30-61), wherein said command instructions comprise instructions to a NoSQL database.
The method of example 29 (as well as subject matter of one or more of any combination of examples 30-62), wherein said command instructions include any combination of one or more of the following: instructions to an operating system (OS), instructions to a data base or instructions to an, SQL interpreter, instructions to a web based language; instructions to a format string interpreter; instructions to a LDAP, interpreter, instructions to an XPath interpreter, instructions to a scripting language, and instructions to a NoSQL database.
An aspect of an embodiment of the present invention provides, but not limited thereto, a system for detecting command injection attacks based on command instructions to be received from a client processor or client data memory.
The system may comprise: a memory unit operative to store software code; and a processor. The processor may be configured to: extract string fragments from the software code to provide extracted signature fragments; receive the client command instructions; convert the received command instructions into command fragments; identify critical parts from the commands fragments; determine if the critical parts are untrusted or trusted by matching with the extracted signature fragments; identify potential attacks upon the condition that a command includes critical parts that are untrusted; and communicate the identification of potential attacks to an output device.
The system of example 64, wherein said output device includes at least one of the following: storage, memory, network, printer or a display.
The system of example 64, wherein the command instructions to be received from the client processor or client data memory includes one or more of the following types: instructions to an operating system (OS), instructions to a data base or instructions to an SQL interpreter, instructions to a web based language; instructions to a format string interpreter; instructions to a LDAP interpreter, XPATH interpreter, instructions to a scripting language, and instructions to a NoSQL database.
The system of example 64, wherein said processor is further configured to: reject said command that includes critical parts that are untrusted.
The system of example 64, wherein said processor is further configured to: remediate said command that includes critical parts that are untrusted;
The system of example 68, wherein said remediation of said identified potential attack command includes providing one or more of the following instructions: terminating the software program, repairing the potential attack command, enabling additional system monitoring, enabling additional analysis, logging that a suspicious command is being attempted, alerting a supervisory remote computer that a breach is being attempted, notifying an administrator of the suspicious command, not issuing the command but returning an error code, not issuing the command but returning a non-error code, issuing a known bad command and returning the resulting error codes, displaying an alert to the end user, or combination thereof.
The system of example 68, said remediation includes altering said identified potential attack command.
The system of example 68, further comprising: a command interpreter module, and wherein said processor is further configured to: transmit said remediated command to said command interpreter module for execution. to generate a command response.
The system of example 68, wherein said processor is further configured to: transmit said remediated command for execution to generate a command response.
The system of example 68, wherein said system further comprise: a command interpreter, and wherein said processor is further configured to: wherein if a command does not include critical parts that are untrusted then the command is deemed safe; and accept said safe command and transmit said safe command to said command interpreter module for execution to generate a command response.
The system of example 64, wherein said processor is further configured to: remediate or reject one or more of said identified potential attack
The system of example 74, further comprising: a command interpreter module, and wherein said processor is further configured to: transmit said remediated command to said command interpreter module for execution. to generate a command response.
The system of example 73, wherein said identifying critical parts includes at least one of the following: parsing, dissecting, lexical analyzing, or tokenizing.
The system of example 64, wherein said extracting comprises a string extracting technique, and said compute processor being configured to: post-process of said extracted signature fragments.
The system of example 77, wherein said post-processing may comprise one or more of the following: removing said extracted signature fragments; adding said extracted signature fragments; modifying said extracted signature fragments; or annotating said extracted signature fragments with additional information.
The system of example 64, wherein said identifying potential attacks includes using annotations.
The system of example 79, wherein said annotations are generated by said post-processing.
The system of example 64, wherein said signature fragments specify patterns.
The system of example 81, wherein said patterns comprise one or more of the following: regular expressions, wild-card specifiers, format string specifiers, context-free grammars, or grammars.
The system of example 64, wherein said memory unit, said processor, and said client processor or client data memory are disposed within a single device.
The system of example 83, wherein said device comprises: smart phone, laptop, computer notebook, iPad, PDA, PC, desktop, tablet, camera, gaming device, or television:
The system of example 64, wherein said memory unit and said processor are remotely located from the client processor or client data memory.
The system of example 85, wherein said processor is in communication with said client processor or client data memory by a wired network or a wireless network.
An aspect of an embodiment of the present invention provides, but not limited thereto, a system for detecting command injection attacks based on command instructions to be received from a client processor or client data memory.
The system may comprise: a memory unit operative to store software code and a processor. The processor may be configured: receive string fragments to provide signature fragments; receive command instructions; convert the received command instructions into command fragments; identify critical parts from the commands fragments; determine untrusted or trusted parts of the command instructions by using the signature fragments; identify potential attacks upon the condition that a command includes critical parts that are untrusted; and communicate the identification of potential attacks to an output device.
The system of example 87, wherein said received string fragments are provided by extracting string fragments from said received software.
The system of example 87, wherein said received string fragments are provided by one or more of any of the following: receiving from the software developer, downloading from a networked resource, distributed with the software, manually specifying string fragments, or static or dynamic analysis of the software code.
The system of example 87, wherein said processor is further configured to: monitor, remediate or reject one or more of said identified potential attack commands.
The system of example 90, wherein said remediation includes altering the identified potential attack command.
The system of example 90, wherein said processor is further configured to: transmit said remediated command to a command interpreter module.
The system of example 92, wherein said processor is further configured to: execute said remediated command in said command interpreter; and generate a command response.
The system of example 90, wherein said processor is further configured to: transmit said remediated command for execution; and generate said executed command as a command response.
The system of example 90, wherein said monitoring, remediation or rejection of said identified potential attack command includes performing one or more of the following actions: terminating the software program, repairing the potential attack command, enabling additional system monitoring, enabling additional analysis, logging that a suspicious command is being attempted, alerting a supervisory remote computer that a breach is being attempted, notifying an administrator of the suspicious command, not issuing the command but returning an error code, not issuing the command but returning a non-error code, issuing a known bad command and returning the resulting error codes, displaying an alert to the end user, or any combination thereof.
The system of example 87, wherein said identifying critical parts includes at least one of the following: parsing, dissecting, scanning, tokenizing, or lexical analysis.
The system of example 87, wherein said processor is further configured to: post-process said received string fragments.
The system of example 97, wherein said post-processing may comprise one or more of the following: removing some of said string fragments; adding additional fragments to said string fragments; modifying said string fragments; or annotating said string fragments with additional information.
The system of example 87, wherein said determining untrusted or trusted parts of the command instructions includes using a trust-verification policy.
The system of example 99, wherein said trust-verification policy includes ensuring that certain critical parts of said command instructions exactly match one signature fragment.
The system of example 87, wherein said identifying potential attacks includes using a trust-verification policy.
The system of example 101, wherein said trust-verification policy includes ensuring that certain critical parts of said command instructions exactly match one signature fragment.
The system of example 98, wherein said identifying potential attacks includes using annotations.
The system of example 87, wherein said identifying potential attacks includes using annotations.
The system of example 103, wherein said annotations are generated by the post-processing.
The system of example 87, wherein said signature fragments specify patterns.
The system of example 106, wherein said patterns comprise one or more of the following: regular expressions, wild-card specifiers, format string specifiers, context-free grammars, or grammars.
The system of example 87, wherein if said identifying potential attacks results in no potential attacks then said command is deemed as safe, thereby defining a safe command.
The system of example 108, wherein said processor is further configured to: transmit said safe command to a command interpreter module.
The system of example 109, wherein said processor is further configured to: execute said safe command; and generate a command response.
The system of example 108, wherein said processor is further configured to accept said safe command.
The system of example 87, wherein said output device includes at least one of the following: storage, memory, network, printer or a display.
The system of example 87, wherein said command instructions comprise instructions to an operating system (OS).
The system of example 87, wherein said command instructions comprise database commands or structured query language (SQL) instructions.
The system of example 87, wherein said command instructions comprise instructions to a format string interpreter.
The system of example 87, wherein said command instructions comprise instruction to an LDAP interpreter.
The system of example 87, wherein said command instructions comprise instruction to an XPATH interpreter.
The system of example 87, wherein said command instructions comprise instructions to a web language.
The system of example 87, wherein said command instructions comprise instructions to a scripting language interpreter.
The system of example 87, wherein said command instructions comprise instructions to a NoSQL database.
The system of example 87, wherein said command instructions include any combination of one or more of the following: instructions to an operating system (OS), instructions to a data base or instructions to an, SQL interpreter, instructions to a web based language; instructions to a format string interpreter; instructions to a LDAP, interpreter, instructions to an XPath interpreter, instructions to a scripting language, and instructions to a NoSQL database.
The system of example 87, wherein said memory unit, said processor, and said client processor or client data memory are disposed within a single device.
The system of example 122, wherein said device comprises: smart phone, laptop, computer notebook, iPad, PDA, PC, desktop, tablet, camera, gaming device, or television:
The system of example 87, wherein said memory unit and said processor are remotely located from the client processor or client data memory.
The system of example 124, wherein said processor is in communication with said client processor or client data memory by a wired network or a wireless network.
An aspect of an embodiment of the present invention provides, but not limited thereto, a non-transitory computer readable medium including instructions executable by a processor for detecting command injection attacks. The instructions may comprise: receiving software code; extracting string fragments from the received software code to provide extracted signature fragments; receiving command instructions; converting the received command instructions into command fragments; identifying critical parts from the commands fragments; determining if the critical parts are untrusted or trusted by matching with the extracted signature fragments; identifying potential attacks upon the condition that a command includes critical parts that are untrusted; and communicating the identification of potential attacks to an output device.
The non-transitory computer readable medium of example 126, wherein said output device includes at least one of the following: storage, memory, network, printer or a display.
The non-transitory computer readable medium of example 126, wherein said instructions further comprise performing any of the steps recited in any one of examples 1-28
An aspect of an embodiment of the present invention provides, but not limited thereto, a non-transitory computer readable medium including instructions executable by a processor for detecting command injection attacks. The instructions may comprise: receiving software code; receiving string fragments to provide signature fragments; receiving command instructions; converting the received command instructions into command fragments; identifying critical parts from the commands fragments; determining untrusted or trusted parts of the command instructions by using the signature fragments; identifying potential attacks upon the condition that a command includes critical parts that are untrusted; and communicating the identification of potential attacks to an output device.
The non-transitory computer readable medium of example 129, wherein said output device includes at least one of the following: storage, memory, network, printer or a display.
The non-transitory computer readable medium of example 129, wherein said instructions further comprise performing any of the steps recited in any one of examples 29-63.
The following patents, applications and publications as listed below and throughout this document are hereby incorporated by reference in their entirety herein. The devices, systems, materials, compositions, networks, computer readable media, and methods of various embodiments of the invention disclosed herein may utilize aspects disclosed in the following references, applications, publications and patents and which are hereby incorporated by reference herein in their entirety (and which are not admitted to be prior art with respect to the present invention by inclusion in this section).
The references listed above, as well as all references cited in the specification, including patents, patent applications, journal articles, and all database entries, are incorporated herein by reference to the extent that they supplement, explain, provide a background for, or teach methodology, techniques, and/or compositions employed herein (and which are not admitted to be prior art with respect to the present invention by inclusion in this section).
In summary, while the present invention has been described with respect to specific embodiments, many modifications, variations, alterations, substitutions, and equivalents will be apparent to those skilled in the art. The present invention is not to be limited in scope by the specific embodiment described herein. Indeed, various modifications of the present invention, in addition to those described herein, will be apparent to those of skill in the art from the foregoing description and accompanying drawings. Accordingly, the invention is to be considered as limited only by the spirit and scope of the following claims, including all modifications and equivalents.
Still other embodiments will become readily apparent to those skilled in this art from reading the above-recited detailed description and drawings of certain exemplary embodiments. It should be understood that numerous variations, modifications, and additional embodiments are possible, and accordingly, all such variations, modifications, and embodiments are to be regarded as being within the spirit and scope of this application. For example, regardless of the content of any portion (e.g., title, field, background, summary, abstract, drawing figure, etc.) of this application, unless clearly specified to the contrary, there is no requirement for the inclusion in any claim herein or of any application claiming priority hereto of any particular described or illustrated activity or element, any particular sequence of such activities, or any particular interrelationship of such elements. Moreover, any activity can be repeated, any activity can be performed by multiple entities, and/or any element can be duplicated. Further, any activity or element can be excluded, the sequence of activities can vary, and/or the interrelationship of elements can vary. Unless clearly specified to the contrary, there is no requirement for any particular described or illustrated activity or element, any particular sequence or such activities, any particular size, speed, material, dimension or frequency, or any particularly interrelationship of such elements. Accordingly, the descriptions and drawings are to be regarded as illustrative in nature, and not as restrictive. Moreover, when any number or range is described herein, unless clearly stated otherwise, that number or range is approximate. When any range is described herein, unless clearly stated otherwise, that range includes all values therein and all sub ranges therein. Any information in any material (e.g., a United States/foreign patent, United States/foreign patent application, book, article, etc.) that has been incorporated by reference herein, is only incorporated by reference to the extent that no conflict exists between such information and the other statements and drawings set forth herein. In the event of such conflict, including a conflict that would render invalid any claim herein or seeking priority hereto, then any such conflicting information in such incorporated by reference material is specifically not incorporated by reference herein.
The present application is a continuation application under 35 U.S.C. § 120 of U.S. patent application Ser. No. 14/442,859, filed May 14, 2015, which is a national stage filing of International Application No. PCT/US2013/070180, filed Nov. 14, 2013, which claims priority under 35 U.S.C. § 119(e) from U.S. Provisional Application Ser. No. 61/726,353, filed Nov. 14, 2012, entitled “Method and System for Securing Applications Against Command Injection Attacks,” the disclosures of which are hereby incorporated by reference herein in their entirety.
This invention was made with government support under Grant No. FA8650-10-C-7025, awarded by the Office of U.S. Air Force. The government has certain rights in the invention.
| Number | Name | Date | Kind |
|---|---|---|---|
| 8806607 | Archer | Aug 2014 | B2 |
| 20060236100 | Baskaran | Oct 2006 | A1 |
| 20110035733 | Horning | Feb 2011 | A1 |
| 20110307451 | El Haddi | Dec 2011 | A1 |
| 20120117644 | Soeder | May 2012 | A1 |
| 20140215631 | Tao | Jul 2014 | A1 |
| Number | Date | Country | |
|---|---|---|---|
| 20190199730 A1 | Jun 2019 | US |
| Number | Date | Country | |
|---|---|---|---|
| 61726353 | Nov 2012 | US |
| Number | Date | Country | |
|---|---|---|---|
| Parent | 14442859 | US | |
| Child | 15492264 | US |
