Abstract: Reliability is one of the essential quality requirements of software systems, while software reliability analysis modeling provides a means for estimating reliability of software, which facilitates effective decision making for quality control. Whether models based on the black-box approach which attracts much attention or models based on the white-box approach which is focused on recently, they all attempt to formulate more accurate models of software reliability. However, these models only consider the user code itself regardless of the runtime environment and ignore the influence of the underlying operating system. In this paper, system calls which the software has to depend on are incorporated, multiple failure modes（failures in kernel mode & failures in user mode） are considered and the behavior of error propagation is modeled. By expanding the classic software reliability model step by step, a more accurate model is developed. Then this model is demonstrated in detail by an example.

Key words: software reliability, system call, failure mode, error propagation, Markov process

摘要： 可靠性是软件系统质量需求中的一个重要指标，而软件可靠性建模为人们提供了一种评估软件可靠性的方法，能够在软件质量控制方面有效地提供决策依据。无论是曾经受到广泛关注的黑盒方法，还是现在大量研究的白盒方法，这些模型都试图越来越精确地刻画软件的可靠性。然而，这些模型都是将软件独立在运行环境之外，只是纯粹地、抽象地考虑软件本身，完全忽略了软件所依赖的操作系统和运行环境对可靠性的影响。将软件运行所依赖的系统调用纳入了考虑范畴，考虑了软件中的多种失效模式（内核态失效、用户态失效），同时对系统中的错误传播行为进行了建模，经过对经典模型逐步扩展，建立了一个更加符合实际情况的软件可靠性模型。通过一个实例详细描述了该模型的如何应用。

关键词: 软件可靠性, 系统调用, 失效模式, 错误传播, 马尔可夫过程