Code reuse is widespread in software development as well as internet of things (IoT) devices. However, code reuse introduces many problems, e.g., software plagiarism and known vulnerabilities. Solving these problems requires extensive manual reverse analysis. Fortunately, binary clone detection can help analysts mitigate manual work by matching reusable code and known parts. However, many binary clone detection methods are not robust to various compiler optimization options and different architectures. While some clone detection methods can be applied across different architectures, they rely on manual features based on human prior knowledge to generate feature vectors for assembly functions and fail to consider the internal associations between features from a semantic perspective. To address this problem, we propose and implement a prototype GeneDiff, a semantic-based representation binary clone detection approach for cross-architectures. GeneDiff utilizes a representation model based on natural language processing (NLP) to generate high-dimensional numeric vectors for each function based on the Valgrind intermediate representation (VEX) representation. This is the first work that translates assembly instructions into an intermediate representation and uses a semantic representation model to implement clone detection for cross-architectures. GeneDiff is robust to various compiler optimization options and different architectures. Compared to approaches using symbolic execution, GeneDiff is significantly more efficient and accurate. The area under the curve (AUC) of the receiver operating characteristic (ROC) of GeneDiff reaches 92.35%, which is considerably higher than the approaches that use symbolic execution. Extensive experiments indicate that GeneDiff can detect similarity with high accuracy even when the code has been compiled with different optimization options and targeted to different architectures. We also use real-world IoT firmware across different architectures as targets, therein proving the practicality of GeneDiff in being able to detect known vulnerabilities.

Hybrid fuzzing is a popular software testing technique that combines random fuzzing with concolic execution. It is widely used in the security domain known for its ability to find deeply hidden vulnerabilities and reach high code coverage. Hybrid fuzzing is based on negating branches in the execution path of a specific input to generate new test cases. However, due to numerous inputs and related branches, it does not show the best of its effectiveness without input and branch selection methods. In this paper, we systematically analyze the branch scheduling problem in the internal attributes of hybrid fuzzing, focusing on the synchronization mechanism. To solve the problems, we propose the Selective Hybrid Fuzzing (SHF) approach with branch scheduling based on binary instrumentation. There are two major parts to the SHF approach: (1) we propose a critical branch selection algorithm to select critical branches by three metrics: hit accuracy, solvability, and complexity; (2) we propose a priority score calculation algorithm to select inputs by the number of critical branches. With the SHF approach, we choose only the branches that can be negated to generate new coverage, instead of repeatedly executing the same branches and generating duplicates of inputs. We implement a hybrid fuzzer called SHFuzz with our SHF approach and compare it with the state-of-the-art hybrid fuzzer QSYM. In the evaluation, SHFuzz outperforms QSYM in 20 real-world applications from the Google Fuzzer Test Suite and other program suites in a 12 h test. On average, SHFuzz achieves 8.40% more code coverage and 100 more unique crashes in each application. Our work also finds existing vulnerabilities 7.85× faster than QSYM. We also find new bugs by SHFuzz, which QSYM fails to find. Our evaluation shows that the selective hybrid fuzzing approach can reduce the number of branches executed in concolic execution, enhancing hybrid fuzzing on code coverage and bug finding capabilities.

Despite greater attention being paid to sensitive-information leakage in the cyberdomain, the sensitive-information problem of the physical domain remains neglected. Anonymous users can easily access the sensitive information of other users, such as transaction information, health status, and addresses, without any advanced technologies. Ideally, secret messages should be protected not only in the cyberdomain but also in the complex physical domain. However, popular steganography schemes only work in the traditional cyberdomain and are useless when physical distortions of messages are unavoidable. This paper first defines the concept of cross-domain steganography, and then proposes EasyStego, a novel cross-domain steganography scheme. EasyStego is based on the use of QR barcodes as carriers; therefore, it is robust to physical distortions in the complex physical domain. Moreover, EasyStego has a large capacity for embeddable secrets and strong scalability in various scenarios. EasyStego uses an AES encryption algorithm to control the permissions of secret messages, which is more effective in reducing the possibility of sensitive-information leakage. Experiments show that EasyStego has perfect robustness and good efficiency. Compared with the best current steganography scheme based on barcodes, EasyStego has greater steganographic capacity and less impact on barcode data. In robustness tests, EasyStego successfully extracts secret messages at different angles and distances. In the case of adding natural textures and importing quantitative error bits, other related steganography techniques fail, whereas EasyStego can extract secret messages with a success rate of nearly 100%.