Search for: All records

Reverse proxy servers play a critical role in optimizing Internet services, offering benefits ranging from load balancing to Denial of Service (DoS) protection. A known shortcoming of such proxies is that the backend server becomes oblivious to the IP address of the client who initiated the connection since all requests are forwarded by the proxy server. For HTTP, this issue is trivially solved by the X-Forwarded-For header, which allows the proxy server to pass to the backend server the IP address of the client that originated the request. Unfortunately, no such equivalent exists for many other protocols. To solve this issue, HAProxy created the PROXY protocol, which communicates client information from a proxy server to a backend server at a lower level in the network stack (Layer 4), making it protocol agnostic. In this work, we are the first to study the use of the PROXY protocol at Internet scale and investigate the security impact of its misconfigurations. We launched a measurement study on the full IPv4 address range and found that, over HTTP, more than 170,000 hosts accept PROXY protocol data from arbitrary sources. We demonstrate how to abuse this protocol to bypass onpath proxies (and their protections) and leak sensitive information from backend infrastructures. We discovered over 10,000 servers that are vulnerable to an access bypass, triggered by injecting a (spoofed) PROXY protocol header. Using this technique, we obtained access to over 500 internal servers providing control over IoT monitoring platforms and smart home automation devices, allowing us to, for example, regulate remote controlled window blinds or control security cameras and alarm systems. Beyond HTTP, we demonstrate how the PROXY protocol can be used to turn over 350 SMTP servers into open relays, enabling an attacker to send arbitrary emails from any email address. In sum, our study exposes how PROXY protocol misconfigurations lead to severe security issues that affect multiple protocols prominently used in the wild. 

Malware detection and classification has been the focus of extensive research over many years. However, less effort has been devoted to developing post-detection systems that identify specific malicious capabilities (or behaviors) in malware. Such systems play a critical part in identifying and mitigating the damage caused by malware attacks. Unfortunately, current methods for identifying malware capabilities involve substantial manual reverse engineering efforts and context switching between multiple tools, which slows down an investigation and gives attackers an advantage. In this paper, we propose DEEPCAPA, an automated postdetection system that uses machine learning to identify potentially malicious capabilities in malware in the form of MITRE ATT&CK techniques. Our system operates on sequences of API calls, extracted from the memory snapshots taken at key points during the (sandboxed) execution of malware. Our results demonstrate that DEEPCAPA can accurately identify malicious capabilities, achieving a precision of 95.80% and a recall of 93.76% across 29 different techniques. 

Fuzzing reliably and efficiently finds bugs in software, including operating system kernels. In general, higher code coverage leads to the discovery of more bugs. This is why most existing kernel fuzzers adopt strategies to generate a series of inputs that attempt to greedily maximize the amount of code that they exercise. However, simply executing code may not be sufficient to reveal bugs that require specific sequences of actions. Synthesizing inputs to trigger such bugs depends on two aspects: (i) the actions the executed code takes, and (ii) the order in which those actions are taken. An action is a high-level operation, such as a heap allocation, that is performed by the executed code and has a specific semantic meaning. ACTOR, our action-guided kernel fuzzing framework, deviates from traditional methods. Instead of focusing on code coverage optimization, our approach generates fuzzer programs (inputs) that leverage our understanding of triggered actions and their temporal relationships. Specifically, we first capture actions that potentially operate on shared data structures at different times. Then, we synthesize programs using those actions as building blocks, guided by bug templates expressed in our domain-specific language. We evaluated ACTOR on four different versions of the Linux kernel, including two well-tested and frequently updated long-term (5.4.206, 5.10.131) versions, a stable (5.19), and the latest (6.2-rc5) release. Our evaluation revealed a total of 41 previously unknown bugs, of which 9 have already been fixed. Interestingly, 15 (36.59%) of them were discovered in less than a day. 

Fuzzing reliably and efficiently finds bugs in software, including operating system kernels. In general, higher code coverage leads to the discovery of more bugs. This is why most existing kernel fuzzers adopt strategies to generate a series of inputs that attempt to greedily maximize the amount of code that they exercise. However, simply executing code may not be sufficient to reveal bugs that require specific sequences of actions. Synthesizing inputs to trigger such bugs depends on two aspects: (i) the actions the executed code takes, and (ii) the order in which those actions are taken. An action is a high-level operation, such as a heap allocation, that is performed by the executed code and has a specific semantic meaning. ACTOR, our action-guided kernel fuzzing framework, deviates from traditional methods. Instead of focusing on code coverage optimization, our approach generates fuzzer programs (inputs) that leverage our understanding of triggered actions and their temporal relationships. Specifically, we first capture actions that potentially operate on shared data structures at different times. Then, we synthesize programs using those actions as building blocks, guided by bug templates expressed in our domain-specific language. We evaluated ACTOR on four different versions of the Linux kernel, including two well-tested and frequently updated long-term (5.4.206, 5.10.131) versions, a stable (5.19), and the latest (6.2-rc5) release. Our evaluation revealed a total of 41 previously unknown bugs, of which 9 have already been fixed. Interestingly, 15 (36.59%) of them were discovered in less than a day. 

Despite remarkable improvements, automatic speech recognition is susceptible to adversarial perturbations. Compared to standard machine learning architectures, these attacks are significantly more challenging, especially since the inputs to a speech recognition system are time series that contain both acoustic and linguistic properties of speech. Extracting all recognition-relevant information requires more complex pipelines and an ensemble of specialized components. Consequently, an attacker needs to consider the entire pipeline. In this paper, we present VENOMAVE, the first training- time poisoning attack against speech recognition. Similar to the predominantly studied evasion attacks, we pursue the same goal: leading the system to an incorrect and attacker-chosen transcription of a target audio waveform. In contrast to evasion attacks, however, we assume that the attacker can only manipulate a small part of the training data without altering the target audio waveform at runtime. We evaluate our attack on two datasets: TIDIGITS and Speech Commands. When poisoning less than 0.17% of the dataset, VENOMAVE achieves attack success rates of more than 80.0%, without access to the victim’s network architecture or hyperparameters. In a more realistic scenario, when the target audio waveform is played over the air in different rooms, VENOMAVE maintains a success rate of up to 73.3%. Finally, VENOMAVE achieves an attack transferability rate of 36.4% between two different model architectures.