Improved Architectures for Secure Intra-process Isolation

Intra-process memory isolation can improve security by enforcing least-privilege at a finer granularity than traditional operating system controls without the context-switch overhead associated with inter-process communication. Because the process has traditionally been a fundamental security boundary, assigning different levels of trust to components within a process is a fundamental change in secure systems design. However, so far there has been little research on the challenges of securely implementing intra-process isolation on top of existing operating system abstractions. We find that frequently-used assumptions in secure system design do not precisely hold under realistic conditions, and that these discrepancies lead to exploitable vulnerabilities. We evaluate two recently-proposed memory isolation systems and show that both are vulnerable to the same generic attacks that break their security model. We then extend a subset of these attacks by applying them to a fully-precise model of control-flow integrity, demonstrating a data-only attack that bypasses both static and dynamic control-flow integrity enforcement by overwriting executable code in-memory even under typical w^x assumptions. From these two results, we propose a set of kernel modifications called Xlock that systemically addresses weaknesses in memory permissions enforcement on Linux, bringing them into line with w^x assumptions. Finally, we present modifications to intra-process isolation systems that preserve efficient userspace component transitions while drastically reducing risk of accidental kernel mismanagement by modeling intra-process components as separate processes from the kernel\u27s perspective. Taken together, these mitigations represent a more robust architecture for efficient and secure intra-process isolation