Defeating memory corruption attacks via pointer taintedness detection

摘要: Most malicious attacks compromise system security through memory corruption exploits. Recently proposed techniques attempt to defeat these by protecting program control data. We have constructed a new class of that can network applications without tampering with any These non-control data represent challenge security. In this paper, we propose an architectural technique both and based on the notion pointer taintedness. A is said be tainted if user input used as value. attack detected whenever value dereferenced during execution. The architecture implemented SimpleScalar processor simulator evaluated using synthetic programs well real-world applications. Our effectively detect attacks, it offers better coverage than current methods. transparent existing programs.