Symbolic Testing of OpenCL Code

We present an effective technique for crosschecking a C or C++ program against an accelerated OpenCL version, as well as a technique for detecting data races in OpenCL programs. Our techniques are implemented in KLEE-CL, a symbolic execution engine based on KLEE and KLEE-FP that supports symbolic reasoning on the equivalence between symbolic values. Our approach is to symbolically model the OpenCL environment using an OpenCL runtime library targeted to symbolic execution. Using this model we are able to run OpenCL programs symbolically, keeping track of memory accesses for the purpose of race detection. We then compare the symbolic result against the plain C or C++ implementation in order to detect mismatches between the two versions. We applied KLEE-CL to the Parboil benchmark suite, the Bullet physics library and the OP2 library, in which we were able to find a total of seven errors: two mismatches between the OpenCL and C implementations, three memory errors, one OpenCL compiler bug and one race condition.

[1]  George S. Avrunin,et al.  Using model checking with symbolic execution to verify parallel numerical programs , 2006, ISSTA '06.

[2]  George C. Necula,et al.  Translation validation for an optimizing compiler , 2000, PLDI '00.

[3]  Matthew B. Dwyer,et al.  Differential symbolic execution , 2008, SIGSOFT '08/FSE-16.

[4]  Michael Boyer.  Automated Dynamic Analysis of CUDA Programs , 2008 .

[5]  Michael Burrows,et al.  Eraser: a dynamic data race detector for multithreaded programs , 1997, TOCS.

[6]  Vikram S. Adve,et al.  LLVM: a compilation framework for lifelong program analysis & transformation , 2004, International Symposium on Code Generation and Optimization, 2004. CGO 2004..

[7]  James C. King,et al.  A new approach to program testing , 1974, Programming Methodology.

[8]  Stavros Tripakis,et al.  Checking Equivalence of SPMD Programs Using Non- Interference , 2010 .

[9]  Paul H. J. Kelly,et al.  Symbolic crosschecking of floating-point and SIMD code , 2011, EuroSys '11.

[10]  Peng Li,et al.  GKLEE: concolic verification and test generation for GPUs , 2012, PPoPP '12.

[11]  George Candea,et al.  Parallel symbolic execution for automated real-world software testing , 2011, EuroSys '11.

[12]  Feng Qin,et al.  GRace: a low-overhead mechanism for detecting data races in GPU programs , 2011, PPoPP '11.

[13]  David L. Dill,et al.  Automatic Formal Verification of Block Cipher Implementations , 2008, 2008 Formal Methods in Computer-Aided Design.

[14]  Paul H. J. Kelly,et al.  Performance analysis of the OP2 framework on many-core architectures , 2011, PERV.

[15]  Jong-Deok Choi,et al.  Hybrid dynamic data race detection , 2003, PPoPP '03.

[16]  Michael F. P. O'Boyle,et al.  A Static Task Partitioning Approach for Heterogeneous Systems Using OpenCL , 2011, CC.

[17]  Michael Burrows,et al.  Eraser: a dynamic data race detector for multi-threaded programs , 1997, TOCS.

[18]  Stephen N. Freund,et al.  FastTrack: efficient and precise dynamic race detection , 2009, PLDI '09.

[19]  Daniel Kroening,et al.  Hardware verification using ANSI-C programs as a reference , 2003, ASP-DAC '03.

[20]  Stavros Tripakis,et al.  Checking Non-Interference in SPMD Programs , 2010 .

[21]  Guodong Li,et al.  Scalable SMT-based verification of GPU kernel functions , 2010, FSE '10.