Sketching concurrent data structures

  • Authors:

  • Armando Solar-Lezama;Christopher Grant Jones;Rastislav Bodik

  • Affiliations:

  • UC Berkeley, Berkeley, CA, USA;UC Berkeley, Berkeley, CA, USA;UC Berkeley, Berkeley, CA, USA

  • Venue:
  • Proceedings of the 2008 ACM SIGPLAN conference on Programming language design and implementation
  • Year:
  • 2008

Quantified Score

Hi-index 0.00

Visualization

Abstract

We describe PSketch, a program synthesizer that helps programmers implement concurrent data structures. The system is based on the concept of sketching, a form of synthesis that allows programmers to express their insight about an implementation as a partial program: a sketch. The synthesizer automatically completes the sketch to produce an implementation that matches a given correctness criteria. PSketch is based on a new counterexample-guided inductive synthesis algorithm (CEGIS) that generalizes the original sketch synthesis algorithm from Solar-Lezama et.al. to cope efficiently with concurrent programs. The new algorithm produces a correct implementation by iteratively generating candidate implementations, running them through a verifier, and if they fail, learning from the counterexample traces to produce a better candidate; converging to a solution in a handful of iterations. PSketch also extends Sketch with higher-level sketching constructs that allow the programmer to express her insight as a "soup" of ingredients from which complicated code fragments must be assembled. Such sketches can be viewed as syntactic descriptions of huge spaces of candidate programs (over 108 candidates for some sketches we resolved). We have used the PSketch system to implement several classes of concurrent data structures, including lock-free queues and concurrent sets with fine-grained locking. We have also sketched some other concurrent objects including a sense-reversing barrier and a protocol for the dining philosophers problem; all these sketches resolved in under an hour.