Ôôóöøøóòòòò Ì Blockinòòõùù Óö Ae Blockin Blockin Blockinòø Êêê Blockinððøý Òòðý××× Óó Óò Blockinùöööòø Çç Blockinø¹çöööòøøø Èöóóööñ×

Con urrent Obje t-Oriented Programs Sridhar Iyer. S. Ramesh. Dept. of Computer S ien e Dept of Computer S ien e IIT Guwahati IIT Bombay Guwahati, INDIA Mumbai, INDIA email : sri iitg.ernet.in ramesh se.iitb.ernet.in Abstra t The obje t-oriented paradigm has been found to be useful for the onstru tion of large and omplex onurrent systems. Rea hability analysis is an important and well-known tool for stati (pre-run-time) analysis of on urrent programs. However, dire t appli ation of traditional rea hability analysis to on urrent obje t-oriented programs has many problems, su h as in omplete analysis for reusable lasses (not safe) and in reased omputational omplexity (not eÆ ient). We have proposed a novel te hnique alled apportioning, for e e tive rea hability analysis of on urrent obje t-oriented programs, that integrates the te hniques of abstra tion ( onsidering a redu ed representation of the system) and partitioning (dividing the system into smaller units). The given program is apportioned into a redu ed version of ea h of its lasses, as well as a redu ed version of the program. The error to be he ked is also de omposed into appropriate sub-properties for he king in the rea hability graphs orresponding to the apportioned program. We have developed a number of apportioning-based algorithms, having di erent degrees of safety and effe tiveness. In this paper, we present the details of one of these algorithms. 1 Introdu tion The obje t-oriented paradigm provides support for the onstru tion of modular and reusable program omponents and hen e is attra tive for the design of large and omplex distributed systems. Rea hability analysis is an important and wellknown stati analysis tool for the pre-run-time veriation of riti al properties in on urrent programs, su h as freedom from deadlo k. The ontrolow graphs of individual pro esses are modi ed to highlight the syn hronization stru ture, abstra ting away other details. Subsequently the omplete statetransition graph of the exe ution, known as the rea hability graph, is onstru ted, thereby modeling the on urrent program as the set of all possible exe ution sequen es ([7℄, [6℄). For properties of the syn hronization stru ture in on urrent programs, rea hability analysis provides the same level of assuran e as formal veri ation. The pra ti al potential of rea hability analysis for analyzing non-trivial on urrent programs has been shown in [9℄, and a simple algorithm for the generation of rea hability graphs is des ribed in gure 1.