Trace-Based Symbolic Analysis for Atomicity Violations

Abstract

Abstract. We propose a symbolic algorithm to accurately predict atomicity violations by analyzing a concrete execution trace of a concurrent program. We use both the execution trace and the program source code to construct a symbolic predictive model, which captures a large set of alternative interleavings of the events of the given trace. We use precise symbolic reasoning with a satisfiability modulo theory (SMT) solver to check the feasible interleavings for atomicity violations. Our algorithm differs from the existing methods in that all reported atomicity violations can appear in the actual program execution; and at the same time the feasible interleavings analyzed by our model are significantly more than other predictive models that guarantee the absence of false alarms.

Soundness of Data Flow Analyses for Weak Memory Models

“…A similar study of the cost of such a fixed point iteration has recently been undertaken by Miné [16], and our experiments confirm his results: we use three sets of benchmarks that were previously used as case studies on the analysis of concurrent software: (1) concurrency bug patterns from the Apache web server as used in [26] (atom001, atom001a, atom002, atom002a, banking/av, banking/no av, banking/some av), (2) the banking and indexer examples from [27] (banking and indexer), and (3) several Linux device drivers together with nondeterministic environments as generated by DDVerify [28]. The detailed results and the source code of all experiments together with our implementation of Rugina and Rinard's points-to analysis is available at http://www.cprover.org/wmm/.…”

Lock Removal for Concurrent Trace Programs

“…The main driving application in this paper is runtime predictive analysis [12,25,5,11,23,29,19], which is a promising method for detecting concurrency bugs by analyzing an execution trace. In other words, even if the given test execution is not erroneous, but if an alternative interleaving of the events of that trace can trigger a failure, runtime predictive analysis will be able to detect it.…”

“…The main scalability bottleneck is step 3 wherein the feasibility check needs to explore all possible interleavings of the trace events. Although the problem in step 3 can be solved by an efficient symbolic analysis [29,19], redundant locks in the trace program can unnecessarily increase the cost of this analysis, since they can lead to a large number of locking constraints that need to be modeled and checked. Our lock removal method can cut down on the number of unnecessary locking constraints, therefore resulting in significant performance improvement in the subsequent analysis.…”

“…We have evaluated the lock removal method in the context of an SMT-based runtime predictive analysis [28,29], to quickly remove the lock statements that are redundant and therefore ease the burden of modeling and checking by the SMT solvers.…”

“…locks) and the shared memory consistency. Please refer to [28,29,26] for more information about the symbolic encoding. Here we assume the sequential consistency (SC) memory model.…”