The verification of cache coherence protocols

Pong, Fong; Dubois, Michel

doi:10.1145/165231.165233

Cited by 37 publications

(47 citation statements)

References 11 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Some concrete examples of verification of cache protocols can be found in [6,22]. Pong and Dubois [24] described general methods that were sound but not complete, as they were based on conservative, inexact abstractions. In [16], it was shown that the PMCP for safety over broadcast protocols [14] is decidable using the general backward reachability procedure of [1].…”

Section: Discussionmentioning

confidence: 99%

“…In the construction, a path x leading to global state s is represented as a tuple of the form (a, A) ∈ S × 2 S , where S is the set of local states of the given cache protocol, that reflects not merely the local states present in s but also takes into account the local transitions that were fired along x to get to s, viz., the history of s along x. The extra historical information, that our construction stores, permits us to reason about safety properties for an arbitrary number of caches in an exact fashion as opposed to the standard abstract graph construction [24] that only takes into account the set of local states present in s and is thus sound but not guaranteed complete. We establish a path correspondence between concrete computations of the original system and paths in the abstract graph which also allows us to automatically generate error traces once an erroneous 'abstract state' is detected.…”

Section: Introductionmentioning

confidence: 99%

See 1 more Smart Citation

Exact and Efficient Verification of Parameterized Cache Coherence Protocols

Kahlon

2003

Lecture Notes in Computer Science

View full text Add to dashboard Cite

Abstract. We propose new, tractably (in some cases provably) efficient algorithmic methods for exact (sound and complete) parameterized reasoning about cache coherence protocols. For reasoning about general snoopy cache protocols, we introduce the guarded broadcast protocols model and show how an abstract history graph construction can be used to reason about safety properties for this framework. Although the worst case size of the abstract history graph can be exponential in the size of the transition diagram of the given protocol, the actual size is small for standard cache protocols as is evidenced by our experimental results. The framework can handle all 8 of the cache protocols in [19] as well as their split-transaction versions. We next identify a framework called initialized broadcast protocols suitable for reasoning about invalidation-based snoopy cache protocols and show how to reduce reasoning about such systems with an arbitrary number of caches to a system with at most 7 caches. This yields a provably polynomial time algorithm for the parameterized verification of invalidation based snoopy protocols. Our results apply to both safety and liveness properties. Finally, we present a methodology for reducing parameterized reasoning about directory based protocols to snoopy protocols, thus leveraging techniques developed for verifying snoopy protocols to directory based ones, which are typically are much harder to reason about. We demonstrate by reducing reasoning about a directory based protocol suggested by German [17] to the ESI snoopy protocol, a modification of the MSI snoopy protocol.

show abstract

Section: Discussionmentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

Exact and Efficient Verification of Parameterized Cache Coherence Protocols

Kahlon

2003

Lecture Notes in Computer Science

View full text Add to dashboard Cite

show abstract

“…The works closest to ours are the methods based on counter abstraction (e.g., [7,24,23]). In particular, verification of liveness properties under fairness is addressed in [23].…”

Section: Discussion and Related Workmentioning

confidence: 99%

“…Since parameterized systems contain process types with large number of behaviorally similar processes (whose behavior follows a local finite state machine or FSM), a natural state space abstraction is to group the processes based on which state of the local FSM they reside in [23,7,24]. Thus, instead of saying "process 1 is in state s, process 2 is in state t and process 3 is in state s" -we simply say "2 processes are in state s and 1 is in state t".…”

Section: Introductionmentioning

confidence: 99%

Fair Model Checking with Process Counter Abstraction

Sun

Liu

Roychoudhury

et al. 2009

Lecture Notes in Computer Science

View full text Add to dashboard Cite

Abstract. Parameterized systems are characterized by the presence of a large (or even unbounded) number of behaviorally similar processes, and they often appear in distributed/concurrent systems. A common state space abstraction for checking parameterized systems involves not keeping track of process identifiers by grouping behaviorally similar processes. Such an abstraction, while useful, conflicts with the notion of fairness. Because process identifiers are lost in the abstraction, it is difficult to ensure fairness (in terms of progress in executions) among the processes. In this work, we study the problem of fair model checking with process counter abstraction. Even without maintaining the process identifiers, our on-the-fly checking algorithm enforces fairness by keeping track of the local states from where actions are enabled / executed within an execution trace. We enhance our home-grown PAT model checker with the technique and show its usability via the automated verification of several real-life protocols.

show abstract

“…Lesens and Saïdi [LS97] combine predicate abstraction with a counting abstraction to verify parameterized networks of processes. A similar 0-1-many abstraction has been studied by Pong and Dubois [PD95] and Pnueli, Xu, and Zuck [PXZ02]. The PAX system [BBLS00] captures parametric systems using the WS1S logic so that finite-state abstractions can be constructed using the MONA tool.…”

Section: Abstracting Parameterized Systemsmentioning

confidence: 97%

Verification by Abstraction

Shankar

2003

Formal Methods at the Crossroads. From Panacea to Foundational Support

View full text Add to dashboard Cite

Abstract. Verification seeks to prove or refute putative properties of a given program. Deductive verification is carried out by constructing a proof that the program satisfies its specification, whereas model checking uses state exploration to find computations where the property fails. Model checking is largely automatic but is effective only for programs defined over small state spaces. Abstraction serves as a bridge between the more general deductive methods for program verification and the restricted but effective state exporation methods used in model checking. In verification by abstraction, deduction is used to construct a finitestate approximation of a program that preserves the property of interest. The resulting abstraction can be explored for offending computations through the use of model checking. We motivate the use of abstraction in verification and survey some of the recent advances.

show abstract

The verification of cache coherence protocols

Cited by 37 publications

References 11 publications

Exact and Efficient Verification of Parameterized Cache Coherence Protocols

Exact and Efficient Verification of Parameterized Cache Coherence Protocols

Fair Model Checking with Process Counter Abstraction

Verification by Abstraction

Contact Info

Product

Resources

About