Search for: All records

The past few years have witnessed a boom of mobile super apps, which are the apps offering multiple services such as e-commerce, e-learning, and e-government via miniapps executed inside. While originally designed for mobile platforms, super apps such as WeChat have also been made available on desktop platforms such as Windows. However, when running on desktop platforms, WeChat experiences differences in some behaviors, which presents opportunities for attacks (e.g., platform fingerprinting attacks). This paper thus aims to systematically identify the potential discrepancies in the APIs of WeChat across platforms and demonstrate how these differences can be exploited by remote attackers or local malicious miniapps. To this end, we present APIDIFF, an automatic tool that generates test cases for each API and identifies execution discrepancies. With APIDIFF, we have identified three sets of discrepant APIs that exhibit existence (109), permission (17), and output (22) discrepancies across platforms and devices, and provided concrete examples of their exploitation. We have responsibly disclosed these vulnerabilities to Tencent and received bug bounties for our findings. These vulnerabilities were ranked as high-severity and some have already been patched. 

An increasing number of Trusted Execution Environment (TEE) is adopting to a variety of commercial products for protecting data security on the cloud. However, TEEs are still exposed to various side-channel vulnerabilities, such as execution order-based, timing-based, and power-based vulnerabilities. While recent hardware is applying various techniques to mitigate order-based and timing-based side-channel vulnerabilities, power-based side-channel attacks remain a concern of hardware security, especially for the confidential computing settings where the server machines are beyond the control of cloud users. In this paper, we present PWRLEAK, an attack framework that exploits AMD’s power reporting interfaces to build power side-channel attacks against AMD Secure Encrypted Virtualization (SEV)-protected VM. We design and implement the attack framework with three general steps: (1) identify the instruction running inside AMD SEV, (2) apply a power interpolator to amplify power consumption, including an emulation-based interpolator for analyzing purposes and a moregeneral interrupt-based interpolator, and (3) infer secrets with various analysis approaches. A case study of using the emulation-based interpolator to infer the whole JPEG images processed by libjpeg demonstrates its ability to help analyze power consumption inside SEV VM. Our end-to-end attacks against Intel’s Integrated Performance Primitives (Intel IPP) library indicates that PWRLEAK can be exploited to infer RSA private keys with over 80% accuracy using the interrupt based interpolator. 

The recent development of Trusted Execution Environment has brought unprecedented opportunities for confidential computing within cloud-based systems. Among various popular cloud business models, serverless computing has gained dominance since its emergence, leading to a high demand for confidential serverless computing services based on trusted enclaves. However, the issue of cold start overhead significantly hinders its performance, as new enclaves need to be created to ensure a clean and verifiable execution environment. In this paper, we propose a novel approach for constructing reusable enclaves that enable rapid enclave reset and robust security with three key enabling techniques: enclave snapshot and rewinding, nested attestation, and multi-layer intra-enclave compartmentalisation. We have built a prototype system for confidential serverless computing, integrating OpenWhisk and a WebAssembly runtime, which significantly reduces the cold start overhead in an end-to-end serverless setting while imposing a reasonable performance impact on standard execution. 

Data processing oriented software, especially machine learning applications, are heavily dependent on standard frameworks/libraries such as TensorFlow and OpenCV. As those frameworks have gained significant popularity, the exploitation of vulnerabilities in the frameworks has become a critical security concern. While software isolation can minimize the impact of exploitation, existing approaches suffer from difficulty analyzing complex program dependencies or excessive overhead, making them ineffective in practice. We propose FreePart, a framework-focused software partitioning technique specialized for data processing applications. It is based on an observation that the execution of a data processing application, including data flows and usage of critical data, is closely related to the invocations of framework APIs. Hence, we conduct a temporal partitioning of the host application’s execution based on the invocations of framework APIs and the data objects used by the APIs. By focusing on data accesses at runtime instead of static program code, it provides effective and practical isolation from the perspective of data. Our evaluation on 23 applications using popular frameworks (e.g., OpenCV, Caffe, PyTorch, and TensorFlow) shows that FreePart is effective against all attacks composed of 18 real-world vulnerabilities with a low overhead (3.68%).