SoD: Synthesizing Device Drivers

[0439017] walid M. TahaWilliam Marsh Rice大学sod：综合设备驱动程序本项目的目标是基于最近在多阶段编程语言方面的工作，开发一个用于设计软件密集型系统的框架。我们的论点是，最初为支持程序优化而开发的分期概念是软件密集型系统设计这门新兴科学中的一项基本技术。我们相信，许多软件密集型系统可以使用语言框架来更有效地设计，该语言框架反映了将高级软件设计与其在特定平台上的实现分开的阶段区别。在早期的工作中，我们帮助开发了多阶段编程的概念基础，包括构建核心演算和支持分期的类型系统，以及开发一种包含分期的现实语言MetaOCaml。在这个项目中，我们打算展示如何使用分期来改进工业中的软件设计过程，在工业中，程序设计和部署实现之间的阶段区别非常明显。我们将重点关注设备驱动程序的设计，因为用于控制设备的部署代码根据设备接口的变幻性和它所连接的硬件平台而有很大的不同。在这种情况下，我们必须扩展分段技术来处理事件驱动的计算、并发性和对资源的严格限制。

Abstract0439017Walid M. TahaWilliam Marsh Rice UniversitySoD: Synthiszizing Device DriversThe goal of this project is develop a framework for designing software-intensive systems based on recent work on multi-stage programming languages. Our thesis is that the concept of staging, originally developed to support program optimization, is a fundamental technique in the emerging science of design of software-intensive systems. We believe that many software-intensive systems can be designed more effectively using a linguistic framework that reflects the staging distinctions separating high-level software designs from their implementations on particular platforms.In earlier work, we helped develop the conceptual foundations of multi-stage programming including the construction of core calculi and type systems that support staging and the development of a realistic language incorporating staging called MetaOCaml. In this project, we intend to show how staging can be used to improve the process of designing software in industry where the phase distinction between program designs and deployed implementations is acute. We will focus on the design of device drivers because the deployed code for controlling a device varies greatly depending on the vagaries of the device interface and the hardware platform to which it is attached. In this context, we must extend staging techniques to handle event-driven computation, concurrency, and strict limitations on resources.