CIVL: A Concurrency Intermediate Verification Language

CIVL：一种并发中间验证语言

• 批准号: 1346769
• 负责人: Stephen Siegel
• 金额: $ 30万
• 依托单位: University of Delaware
• 依托单位国家: 美国
• 项目类别: Standard Grant
• 财政年份: 2013
• 资助国家: 美国
• 起止时间: 2013-08-01 至 2016-07-31
• 项目状态: 已结题
• 来源: https://www.nsf.gov/awardsearch/showAward?AWD_ID=1346769&HistoricalAwards=false
• 关键词: CIVL; Concurrency; Intermediate; Verification; Language

Parallel programming has become increasingly common with the advent of"multi-core" computer chips that pack many processors into a singlechip. Most standard laptop and desktop computers now come with anywherefrom two to thirty-two such cores. Such chips offer unprecedented computing power, but at a price: toextract their full potential, programmers writing software for the newchips must use special programming language "libraries" in order tospecify how the program is to do two (or more) things at once. Onepopular such library is called OpenMP. These so-called parallelprograms are notoriously difficult to get right, and many containundetected defects ("bugs") that can cause the program to crash orproduce incorrect results.The CIVL project has the potential to dramatically reduce the effortrequired to develop new static analysis tools. The key idea is aunified language for describing parallel programs, whether these useMPI, OpenMP, or both. One tool translates the original program intothe new language, also called CIVL. Other tools perform staticanalysis on the CIVL program. Eventually, many other parallellibraries will be added to the new system. The advantage of thisapproach is that the designer of a new static analysis tool need onlydesign the tool for a single language---CIVL---but then gets a toolthat works on all the source libraries "for free". Similarly, when anew parallel library comes along, by developing a translator from itto CIVL, one can immeidately reap the benefits of all the staticanalysis tools. The researchers expect that the resulting platformwill make it much easier to develop correct parallel programs, nomatter how that parallelism is expressed.

随着“多核”计算机芯片的出现，将许多处理器纳入单人货币时，并行编程变得越来越普遍。现在，大多数标准的笔记本电脑和台式计算机都从两到三十二个这样的核心中都带有。这样的芯片提供了前所未有的计算能力，但是以一定的代价：为NewChips编写软件的程序员必须使用特殊的编程语言“库”，以便该程序如何一次（或更多）进行两项（或更多）。一库称为OpenMP。众所周知，这些所谓的并行程序很难正确地正确，并且许多可以导致程序崩溃或生产不正确结果的缺陷（“错误”）。CIVL项目有可能大大减少实体式以开发新的静态分析工具。关键想法是用于描述并行程序的aunatied语言，无论是这些USEMPI，OpenMP还是两者兼而有之。一种工具翻译了原始程序中的新语言，也称为CIVL。其他工具对CIVL程序进行静态分析。最终，将添加许多其他等级元素。这种申请的优点在于，新的静态分析工具的设计师只需要使用单个语言的工具来设计工具--- civl ---但然后获得工具，可以在所有源库中使用“免费”。同样，当新的平行库出现时，通过从ITTO CIVL开发翻译人员，就可以毫无疑问地从所有静态分析工具中获得好处。研究人员预计，由此产生的平台将使制定正确的并行程序变得更加容易，即如何表达并行性。