设计鲁棒的API监控解决方案
Designing Robust API Monitoring Solutions
AI总结：
针对现实软件栈耦合下API监控兼顾准确性、可靠性、透明性的难题，明确六大挑战与设计要点，提出基于硬件辅助虚拟化（首款通用用户空间追踪器）和动态二进制翻译的两种实现，开源SNIPER系统。
摘要：
追踪程序执行的库调用与系统调用序列，对表征其与周边环境的交互乃至最终的语义特征极具价值。由于现实软件栈的复杂耦合，在兼顾准确性、可靠性与透明性的前提下完成这项任务，难度远超预期。为有效平衡这些维度，我们明确了API监控方案需攻克的六大挑战，梳理了可落地的设计要点，并分享了我们为软件安全研究构建API追踪器的实践经验。我们详细介绍了两种实现变体：分别基于硬件辅助虚拟化（打造了首款此类通用用户空间追踪器）和动态二进制翻译，二者均可实现稳健的API监控。我们已将SNIPER系统开源。
英文摘要：
Tracing the sequence of library and system calls that a program makes is very helpful in the characterization of its interactions with the surrounding environment and ultimately of its semantics. Due to entanglements of real-world software stacks, accomplishing this task can be surprisingly challenging as we take accuracy, reliability, and transparency into the equation. To manage these dimensions effectively, we identify six challenges that API monitoring solutions should overcome and outline actionable design points for them, reporting insights from our experience in building API tracers for software security research. We detail two implementation variants, based on hardware-assisted virtualization (realizing the first general-purpose user-space tracer of this kind) and on dynamic binary translation, that achieve API monitoring robustly. We share our SNIPER system as open source.
机构、总结、中文摘要均由AI分析而来，可能存在错误，请以论文正文为准。
