Probabilistic Calling Context

Size: px
Start display at page:

Download "Probabilistic Calling Context"

Transcription

1 Probabilistic Calling Context Michael D. Bond Kathryn S. McKinley University of Texas at Austin Why Context Sensitivity? Static program location not enough at com.mckoi.db.jdbcserver.jdbcinterface.execquery():213 1

2 Why Context Sensitivity? Static program location not enough at com.mckoi.db.jdbcserver.jdbcinterface.execquery():213 at com.mckoi.db.jdbc.mconnection.executequery():348 at com.mckoi.db.jdbc.mstatement.executequery():11 at com.mckoi.db.jdbc.mstatement.executequery():127 at Test.main():48 Why Context Sensitivity? Static program location not enough at com.mckoi.db.jdbcserver.jdbcinterface.execquery():213 at com.mckoi.db.jdbc.mconnection.executequery():348 at com.mckoi.db.jdbc.mstatement.executequery():11 at com.mckoi.db.jdbc.mstatement.executequery():127 at Test.main():48 Motivated by Complex programs Small methods Virtual dispatch 2

3 Why Context Sensitivity? Static program location not enough at com.mckoi.db.jdbcserver.jdbcinterface.execquery():213 at com.mckoi.db.jdbc.mconnection.executequery():348 at com.mckoi.db.jdbc.mstatement.executequery():11 at com.mckoi.db.jdbc.mstatement.executequery():127 at Test.main():48 Motivated by Complex programs Small methods Virtual dispatch C/Fortran method call return Java/C# method return call Context Is Nontrivial Program antlr bloat chart eclipse fop hsqldb jython luindex lusearch pmd xalan Call sites 4,184 3,36 2,335 9,611 2, , ,89 1,53 API calls Distinct contexts 128,627 6,947 22,63 226,2 37,71 16,5 628,48 12, ,18 17,95 3

4 Example: Residual Testing Does behavior occur at production time that did not occur at testing time? class SimpleWindow { close() { class EditorWindow { close() { Example: Residual Testing Does behavior occur at production time that did not occur at testing time? autoupdate() { class SimpleWindow { close() { for all windows w w.close(); inputhandler() { class EditorWindow { case CLICK_EXIT: close() { w.checkunsaved(); w.close(); 4

5 Example: Residual Testing Does behavior occur at production time that did not occur at testing time? autoupdate() { class SimpleWindow { close() { for all windows w w.close(); inputhandler() { case CLICK_EXIT: w.checkunsaved(); w.close(); class EditorWindow { close() { Bug! Example: Residual Testing Does behavior occur at production time that did not occur at testing time? autoupdate() { class SimpleWindow { close() { for all windows w New behavior indicates bugs w.close(); inputhandler() { class EditorWindow { find new behavior case CLICK_EXIT: close() { w.checkunsaved(); Bug! w.close(); Context sensitivity helps 5

6 Two-Phase Dynamic Analyses Training Production Behavior observed New or anomalous behavior detected Two-Phase Dynamic Analyses Residual testing What behavior occurs at production time that did not occur at testing time? [Vaswani et al. 7] Anomaly-based bug detection What new behavior occurs during a buggy program run? [Hangal & Lam 2] Anomaly-based intrusion detection Does a program exhibit anomalous behavior? [Inoue 5] Training Production Behavior observed New or anomalous behavior detected 6

7 Probabilistic Calling Context Adds context sensitivity to dynamic analyses Maintains value representing context Unique with high probability New value new context walk stack High accuracy: <.1% false negatives Low overhead: 3% overhead, -8% for clients Training Production Behavior observed New or anomalous behavior detected Outline Introduction Previous approaches Maintaining the PCC value Evaluation Accuracy Performance 7

8 Previous Approaches Tracking context [Ammons et al. 97] [Spivey 4] Maintain CCT position at each call/return Walking the stack [Nethercote & Seward 7] Path profiling [Ball & Larus 96] [Melski & Reps 99] Call graphs large path explosion Virtual dispatch complicates instrumentation Previous Approaches Tracking context [Ammons et al. 97] [Spivey 4] Maintain CCT position at each call/return Walking the stack [Nethercote & Seward 7] Path profiling [Ball & Larus 96] [Melski & Reps 99] Call graphs large path explosion Virtual dispatch complicates instrumentation Sampling [Zhuang et al. 6] Sacrifices coverage for low overhead 8

9 Outline Introduction Previous approaches Maintaining the PCC value Evaluation Accuracy Performance PCC Function f ( V, cs ) V is PCC value cs is call site ID 9

10 PCC Function f ( V, cs ) V is PCC value cs is call site ID cs 1 V f ( V, cs 1 ) cs V f ( V, cs 2 ) 2 V V saved V V saved PCC Function f ( V, cs ) 3V + cs (mod 2 32 ) V is PCC value cs is call site ID 1

11 PCC Function f ( V, cs ) 3V + cs (mod 2 32 ) Motivated by MPI datatype hashing [Langou et al. 5] [Gropp ] Cheap to compute Desirable properties: Non-commutative Composition efficient to compute Differentiating Similar Contexts A V 3V + cs 1 V 3V + cs 2 B C V 3V + cs 2 A B V 3V + cs 1 C A() B() B() A() 11

12 Differentiating Similar Contexts A V 3V + cs 1 V 3V + cs 2 B C V 3V + cs 2 A B V 3V + cs 1 C Non-commutative f ( f (V, cs 1 ), cs 2 ) f ( f (V, cs 2 ), cs 1 ) Efficiency at Inlined Calls A V 3V + cs 1 B V 3V + cs 2 C 12

13 Efficiency at Inlined Calls A V 3V + cs 1 A B V 3V + cs 2 C B V 3 ( 3V + cs 1 ) + cs 2 C Efficiency at Inlined Calls A V 3V + cs 1 A B V 3V + cs 2 C B V 9V + 3cs 1 + cs 2 C 13

14 Efficiency at Inlined Calls A V 3V + cs 1 A B V 3V + cs 2 C B V 9V + 3cs 1 + cs 2 C Composition efficient to compute n n i f ( V, cs ) = 3 V + 3 cs i i i Outline Introduction Previous approaches Maintaining the PCC value Evaluation Methodology Evaluating potential clients Accuracy Performance 14

15 Methodology Implementation in Jikes RVM Available on Jikes RVM Research Archive Deterministic calling context profiling Maintains CCT node at each call & return Benchmarks: DaCapo, SPEC JBB2, SPEC JVM98 Platform: 3.6 GHz Pentium 4 w/linux How Clients Use PCC Record values New value new context walk stack Training Production Behavior observed New or anomalous behavior detected 15

16 Evaluating Potential Clients Record values Training Global hash table Check values (no new values) Production Behavior observed New or anomalous behavior detected Evaluating Potential Clients Memory overhead: proportional to contexts Record values Training Global hash table Check values (no new values) Production Behavior observed New or anomalous behavior detected 16

17 Evaluating Potential Clients Anomaly-based intrusion detection Check PCC value at system calls (Network, I/O, OS) Residual testing Check PCC value at Java API calls (calls to java.*) Upper bound Check PCC value at all calls Ideal Accuracy PCC maps context to value New PCC value new context Familiar PCC value probably familiar context 17

18 Ideal Accuracy PCC maps context to value New PCC value new context Familiar PCC value probably familiar context Expected conflicts (false negatives) Distinct contexts 1, 1,, 1,, 1,, 1,,, 1,,, 32-bit values 1 (.%) 116 (.%) 11,632 (.1%) 1,155,17 (1.2%) 17,882,641 (1.8%) 6,123,623,65 (61.2%) 64-bit values (.%) (.%) (.%) (.%) (.%) 3 (.%) Ideal Accuracy PCC maps context to value New PCC value new context Familiar PCC value probably familiar context Expected conflicts (false negatives) Distinct contexts 32-bit values 64-bit values 1, 1,, API calls 1 (.%) 116 (.%) (.%) (.%) 1,, 11,632 (.1%) (.%) 1,, 1,,, 1,,, 1,155,17 (1.2%) 17,882,641 (1.8%) 6,123,623,65 (61.2%) (.%) (.%) 3 (.%) 18

19 Ideal Accuracy PCC maps context to value New PCC value new context Familiar PCC value probably familiar context Expected conflicts (false negatives) Distinct contexts 32-bit values 1, 1 (.%) 1,, 116 (.%) 1,, 11,632 (.1%) 64-bit values (.%) (.%) All calls (.%) 1,, 1,155,17 (1.2%) (.%) 1,,, 17,882,641 (1.8%) (.%) 1,,, 6,123,623,65 (61.2%) 3 (.%) Ideal Accuracy PCC maps context to value New PCC value new context Familiar PCC value probably familiar context Expected conflicts (false negatives) Distinct contexts 32-bit values 64-bit values 1, 1 (.%) (.%) 1,, 116 (.%) (.%) Near-perfect 1,, 11,632 (.1%) accuracy (.%) 1,, 1,155,17 (1.2%) (.%) 1,,, 17,882,641 (1.8%) (.%) 1,,, 6,123,623,65 (61.2%) 3 (.%) 19

20 2 PCC s Accuracy 17,95 2,15,838, ,895 xalan ,18 537,17, ,45 pmd ,511, lusearch 12,556 39,733, ,615 luindex ,48 543,845,772 4,289 5,929 jython 16,5 81,161, hsqldb 37,71 9,918, fop 5 226,2 132,57, ,11 eclipse 4 22,63 258,891, chart 4 6,947 1,159,281, bloat 3 128,627 24,422,13 1, ,49 antlr Conf. Distinct Dynamic Conf. Distinct Dynamic Program Java API calls System calls PCC s Accuracy 17,95 2,15,838, ,895 xalan ,18 537,17, ,45 pmd ,511, lusearch 12,556 39,733, ,615 luindex ,48 543,845,772 4,289 5,929 jython 16,5 81,161, hsqldb 37,71 9,918, fop 5 226,2 132,57, ,11 eclipse 4 22,63 258,891, chart 4 6,947 1,159,281, bloat 3 128,627 24,422,13 1, ,49 antlr Conf. Distinct Dynamic Conf. Distinct Dynamic Program Java API calls System calls

21 PCC s Accuracy Program antlr bloat chart eclipse fop hsqldb jython luindex lusearch pmd xalan Dynamic 49,363,211 6,276,446,59 98,459,469 1,266,81,54 44,2, ,68,667 5,326,949,158 74,53,14 1,439,34,336 2,726,876,957 1,83,858,546 All calls Distinct 1,6,578 1,98,25 845,432 4,815,91 174,955 11,795 3,859, ,21 6,39 8,43,96 163,25 Conf , , ,653 6 PCC s Execution Time Overhead DaCapo SPEC All 6 5 Overhead (%) PCC 3% PCC + system calls 21

22 PCC s Execution Time Overhead DaCapo SPEC All 6 5 Overhead (%) % PCC PCC + system calls PCC + API calls PCC + all calls Summary PCC maintains calling context value New value indicates new behavior Low overhead Maintaining PCC value adds 3% Checking PCC value -8% Memory overhead proportional to contexts High accuracy Less than.1% false negative rate PCC adds context sensitivity to clients that detect anomalous behavior 22

23 Summary Thank you! PCC maintains calling context value New value indicates new behavior Low overhead Maintaining PCC value adds 3% Checking PCC value -8% Memory overhead proportional to contexts High accuracy Less than.1% false negative rate PCC adds context sensitivity to clients that detect anomalous behavior Extra slides 23

24 Context Sensitivity Mostly Unused Do paths capture enough behavior? C/Fortran method call call return Java/C# method return 24

Program Calling Context. Motivation

Program Calling Context. Motivation Program alling ontext Motivation alling context enhances program understanding and dynamic analyses by providing a rich representation of program location ollecting calling context can be expensive The

More information

Towards Parallel, Scalable VM Services

Towards Parallel, Scalable VM Services Towards Parallel, Scalable VM Services Kathryn S McKinley The University of Texas at Austin Kathryn McKinley Towards Parallel, Scalable VM Services 1 20 th Century Simplistic Hardware View Faster Processors

More information

How s the Parallel Computing Revolution Going? Towards Parallel, Scalable VM Services

How s the Parallel Computing Revolution Going? Towards Parallel, Scalable VM Services How s the Parallel Computing Revolution Going? Towards Parallel, Scalable VM Services Kathryn S McKinley The University of Texas at Austin Kathryn McKinley Towards Parallel, Scalable VM Services 1 20 th

More information

Lu Fang, University of California, Irvine Liang Dou, East China Normal University Harry Xu, University of California, Irvine

Lu Fang, University of California, Irvine Liang Dou, East China Normal University Harry Xu, University of California, Irvine Lu Fang, University of California, Irvine Liang Dou, East China Normal University Harry Xu, University of California, Irvine 2015-07-09 Inefficient code regions [G. Jin et al. PLDI 2012] Inefficient code

More information

A Dynamic Evaluation of the Precision of Static Heap Abstractions

A Dynamic Evaluation of the Precision of Static Heap Abstractions A Dynamic Evaluation of the Precision of Static Heap Abstractions OOSPLA - Reno, NV October 20, 2010 Percy Liang Omer Tripp Mayur Naik Mooly Sagiv UC Berkeley Tel-Aviv Univ. Intel Labs Berkeley Tel-Aviv

More information

Efficient Runtime Tracking of Allocation Sites in Java

Efficient Runtime Tracking of Allocation Sites in Java Efficient Runtime Tracking of Allocation Sites in Java Rei Odaira, Kazunori Ogata, Kiyokuni Kawachiya, Tamiya Onodera, Toshio Nakatani IBM Research - Tokyo Why Do You Need Allocation Site Information?

More information

Bugs in Deployed Software

Bugs in Deployed Software Bell: Bit-Encoding Online Memory Leak Detection Michael D. Bond Kathryn S. McKinley University of Texas at Austin Bugs in Deployed Software Humans rely on software for critical tasks Bugs are costly &

More information

A Trace-based Java JIT Compiler Retrofitted from a Method-based Compiler

A Trace-based Java JIT Compiler Retrofitted from a Method-based Compiler A Trace-based Java JIT Compiler Retrofitted from a Method-based Compiler Hiroshi Inoue, Hiroshige Hayashizaki, Peng Wu and Toshio Nakatani IBM Research Tokyo IBM Research T.J. Watson Research Center April

More information

Identifying the Sources of Cache Misses in Java Programs Without Relying on Hardware Counters. Hiroshi Inoue and Toshio Nakatani IBM Research - Tokyo

Identifying the Sources of Cache Misses in Java Programs Without Relying on Hardware Counters. Hiroshi Inoue and Toshio Nakatani IBM Research - Tokyo Identifying the Sources of Cache Misses in Java Programs Without Relying on Hardware Counters Hiroshi Inoue and Toshio Nakatani IBM Research - Tokyo June 15, 2012 ISMM 2012 at Beijing, China Motivation

More information

Adaptive Multi-Level Compilation in a Trace-based Java JIT Compiler

Adaptive Multi-Level Compilation in a Trace-based Java JIT Compiler Adaptive Multi-Level Compilation in a Trace-based Java JIT Compiler Hiroshi Inoue, Hiroshige Hayashizaki, Peng Wu and Toshio Nakatani IBM Research Tokyo IBM Research T.J. Watson Research Center October

More information

Lightweight Data Race Detection for Production Runs

Lightweight Data Race Detection for Production Runs Lightweight Data Race Detection for Production Runs Swarnendu Biswas, UT Austin Man Cao, Ohio State University Minjia Zhang, Microsoft Research Michael D. Bond, Ohio State University Benjamin P. Wood,

More information

Efficient Context Sensitivity for Dynamic Analyses via Calling Context Uptrees and Customized Memory Management

Efficient Context Sensitivity for Dynamic Analyses via Calling Context Uptrees and Customized Memory Management to Reuse * * Evaluated * OOPSLA * Artifact * AEC Efficient Context Sensitivity for Dynamic Analyses via Calling Context Uptrees and Customized Memory Management Jipeng Huang Michael D. Bond Ohio State

More information

Toward Efficient Strong Memory Model Support for the Java Platform via Hybrid Synchronization

Toward Efficient Strong Memory Model Support for the Java Platform via Hybrid Synchronization Toward Efficient Strong Memory Model Support for the Java Platform via Hybrid Synchronization Aritra Sengupta, Man Cao, Michael D. Bond and Milind Kulkarni 1 PPPJ 2015, Melbourne, Florida, USA Programming

More information

Chronicler: Lightweight Recording to Reproduce Field Failures

Chronicler: Lightweight Recording to Reproduce Field Failures Chronicler: Lightweight Recording to Reproduce Field Failures Jonathan Bell, Nikhil Sarda and Gail Kaiser Columbia University, New York, NY USA What happens when software crashes? Stack Traces Aren

More information

Phosphor: Illuminating Dynamic. Data Flow in Commodity JVMs

Phosphor: Illuminating Dynamic. Data Flow in Commodity JVMs Phosphor: Illuminating Dynamic Fork me on Github Data Flow in Commodity JVMs Jonathan Bell and Gail Kaiser Columbia University, New York, NY USA Dynamic Data Flow Analysis: Taint Tracking Output that is

More information

Elfen Scheduling. Fine-Grain Principled Borrowing from Latency-Critical Workloads using Simultaneous Multithreading (SMT) Xi Yang

Elfen Scheduling. Fine-Grain Principled Borrowing from Latency-Critical Workloads using Simultaneous Multithreading (SMT) Xi Yang Elfen Scheduling Fine-Grain Principled Borrowing from Latency-Critical Workloads using Simultaneous Multithreading (SMT) Xi Yang Australian National University Stephen M Blackburn Australian National University

More information

GC Assertions: Using the Garbage Collector to Check Heap Properties

GC Assertions: Using the Garbage Collector to Check Heap Properties GC Assertions: Using the Garbage Collector to Check Heap Properties Edward E. Aftandilian Samuel Z. Guyer Department of Computer Science Tufts University {eaftan,sguyer}@cs.tufts.edu Abstract This paper

More information

Phase-based Adaptive Recompilation in a JVM

Phase-based Adaptive Recompilation in a JVM Phase-based Adaptive Recompilation in a JVM Dayong Gu Clark Verbrugge Sable Research Group, School of Computer Science McGill University, Montréal, Canada {dgu1, clump}@cs.mcgill.ca April 7, 2008 Sable

More information

Free-Me: A Static Analysis for Automatic Individual Object Reclamation

Free-Me: A Static Analysis for Automatic Individual Object Reclamation Free-Me: A Static Analysis for Automatic Individual Object Reclamation Samuel Z. Guyer, Kathryn McKinley, Daniel Frampton Presented by: Jason VanFickell Thanks to Dimitris Prountzos for slides adapted

More information

Jinn: Synthesizing Dynamic Bug Detectors for Foreign Language Interfaces

Jinn: Synthesizing Dynamic Bug Detectors for Foreign Language Interfaces Jinn: Synthesizing Dynamic Bug Detectors for Foreign Language Interfaces Byeongcheol Lee Ben Wiedermann Mar>n Hirzel Robert Grimm Kathryn S. McKinley B. Lee, B. Wiedermann, M. Hirzel, R. Grimm, and K.

More information

Tolerating Memory Leaks

Tolerating Memory Leaks Tolerating Memory Leaks UT Austin Technical Report TR-07-64 December 7, 2007 Michael D. Bond Kathryn S. McKinley Department of Computer Sciences The University of Texas at Austin {mikebond,mckinley}@cs.utexas.edu

More information

Understanding Parallelism-Inhibiting Dependences in Sequential Java

Understanding Parallelism-Inhibiting Dependences in Sequential Java Understanding Parallelism-Inhibiting Dependences in Sequential Java Programs Atanas(Nasko) Rountev Kevin Van Valkenburgh Dacong Yan P. Sadayappan Ohio State University Overview and Motivation Multi-core

More information

Trace-based JIT Compilation

Trace-based JIT Compilation Trace-based JIT Compilation Hiroshi Inoue, IBM Research - Tokyo 1 Trace JIT vs. Method JIT https://twitter.com/yukihiro_matz/status/533775624486133762 2 Background: Trace-based Compilation Using a Trace,

More information

Z-Rays: Divide Arrays and Conquer Speed and Flexibility

Z-Rays: Divide Arrays and Conquer Speed and Flexibility Z-Rays: Divide Arrays and Conquer Speed and Flexibility Jennifer B. Sartor Stephen M. Blackburn Daniel Frampton Martin Hirzel Kathryn S. McKinley University of Texas at Austin Australian National University

More information

Uncovering Performance Problems in Java Applications with Reference Propagation Profiling

Uncovering Performance Problems in Java Applications with Reference Propagation Profiling Uncovering Performance Problems in Java Applications with Reference Propagation Profiling Dacong Yan 1 Guoqing Xu 2 Atanas Rountev 1 1 Department of Computer Science and Engineering, Ohio State University

More information

Towards Garbage Collection Modeling

Towards Garbage Collection Modeling Towards Garbage Collection Modeling Peter Libič Petr Tůma Department of Distributed and Dependable Systems Faculty of Mathematics and Physics, Charles University Malostranské nám. 25, 118 00 Prague, Czech

More information

The DaCapo Benchmarks: Java Benchmarking Development and Analysis

The DaCapo Benchmarks: Java Benchmarking Development and Analysis The DaCapo Benchmarks: Java Benchmarking Development and Analysis Stephen M Blackburn αβ, Robin Garner β, Chris Hoffmann γ, Asjad M Khan γ, Kathryn S McKinley δ, Rotem Bentzur ε, Amer Diwan ζ, Daniel Feinberg

More information

Averroes: Letting go of the library!

Averroes: Letting go of the library! Workshop on WALA - PLDI 15 June 13th, 2015 public class HelloWorld { public static void main(string[] args) { System.out.println("Hello, World!"); 2 public class HelloWorld { public static void main(string[]

More information

Continuous Object Access Profiling and Optimizations to Overcome the Memory Wall and Bloat

Continuous Object Access Profiling and Optimizations to Overcome the Memory Wall and Bloat Continuous Object Access Profiling and Optimizations to Overcome the Memory Wall and Bloat Rei Odaira, Toshio Nakatani IBM Research Tokyo ASPLOS 2012 March 5, 2012 Many Wasteful Objects Hurt Performance.

More information

Legato: Bounded Region Serializability Using Commodity Hardware Transactional Memory. Aritra Sengupta Man Cao Michael D. Bond and Milind Kulkarni

Legato: Bounded Region Serializability Using Commodity Hardware Transactional Memory. Aritra Sengupta Man Cao Michael D. Bond and Milind Kulkarni Legato: Bounded Region Serializability Using Commodity Hardware al Memory Aritra Sengupta Man Cao Michael D. Bond and Milind Kulkarni 1 Programming Language Semantics? Data Races C++ no guarantee of semantics

More information

PERFBLOWER: Quickly Detecting Memory-Related Performance Problems via Amplification

PERFBLOWER: Quickly Detecting Memory-Related Performance Problems via Amplification PERFBLOWER: Quickly Detecting Memory-Related Performance Problems via Amplification Lu Fang 1, Liang Dou 2, and Guoqing Xu 1 1 University of California, Irvine, USA {lfang3, guoqingx}@ics.uci.edu 2 East

More information

Immix: A Mark-Region Garbage Collector with Space Efficiency, Fast Collection, and Mutator Performance

Immix: A Mark-Region Garbage Collector with Space Efficiency, Fast Collection, and Mutator Performance Immix: A Mark-Region Garbage Collector with Space Efficiency, Fast Collection, and Mutator Performance Stephen M. Blackburn Australian National University Steve.Blackburn@anu.edu.au Kathryn S. McKinley

More information

Java Performance Evaluation through Rigorous Replay Compilation

Java Performance Evaluation through Rigorous Replay Compilation Java Performance Evaluation through Rigorous Replay Compilation Andy Georges Lieven Eeckhout Dries Buytaert Department Electronics and Information Systems, Ghent University, Belgium {ageorges,leeckhou}@elis.ugent.be,

More information

1 Scalable Runtime Bloat Detection Using Abstract Dynamic Slicing

1 Scalable Runtime Bloat Detection Using Abstract Dynamic Slicing 1 Scalable Runtime Bloat Detection Using Abstract Dynamic Slicing GUOQING XU, University of California, Irvine NICK MITCHELL, IBM T. J. Watson Research Center MATTHEW ARNOLD, IBM T. J. Watson Research

More information

NDSeq: Runtime Checking for Nondeterministic Sequential Specs of Parallel Correctness

NDSeq: Runtime Checking for Nondeterministic Sequential Specs of Parallel Correctness EECS Electrical Engineering and Computer Sciences P A R A L L E L C O M P U T I N G L A B O R A T O R Y NDSeq: Runtime Checking for Nondeterministic Sequential Specs of Parallel Correctness Jacob Burnim,

More information

Adaptive Optimization using Hardware Performance Monitors. Master Thesis by Mathias Payer

Adaptive Optimization using Hardware Performance Monitors. Master Thesis by Mathias Payer Adaptive Optimization using Hardware Performance Monitors Master Thesis by Mathias Payer Supervising Professor: Thomas Gross Supervising Assistant: Florian Schneider Adaptive Optimization using HPM 1/21

More information

Hybrid Static-Dynamic Analysis for Statically Bounded Region Serializability

Hybrid Static-Dynamic Analysis for Statically Bounded Region Serializability Hybrid Static-Dynamic Analysis for Statically Bounded Region Serializability Aritra Sengupta, Swarnendu Biswas, Minjia Zhang, Michael D. Bond and Milind Kulkarni ASPLOS 2015, ISTANBUL, TURKEY Programming

More information

IBM Research - Tokyo 数理 計算科学特論 C プログラミング言語処理系の最先端実装技術. Trace Compilation IBM Corporation

IBM Research - Tokyo 数理 計算科学特論 C プログラミング言語処理系の最先端実装技術. Trace Compilation IBM Corporation 数理 計算科学特論 C プログラミング言語処理系の最先端実装技術 Trace Compilation Trace JIT vs. Method JIT https://twitter.com/yukihiro_matz/status/533775624486133762 2 Background: Trace-based Compilation Using a Trace, a hot path identified

More information

DoubleChecker: Efficient Sound and Precise Atomicity Checking

DoubleChecker: Efficient Sound and Precise Atomicity Checking DoubleChecker: Efficient Sound and Precise Atomicity Checking Swarnendu Biswas, Jipeng Huang, Aritra Sengupta, and Michael D. Bond The Ohio State University PLDI 2014 Impact of Concurrency Bugs Impact

More information

New Algorithms for Static Analysis via Dyck Reachability

New Algorithms for Static Analysis via Dyck Reachability New Algorithms for Static Analysis via Dyck Reachability Andreas Pavlogiannis 4th Inria/EPFL Workshop February 15, 2018 A. Pavlogiannis New Algorithms for Static Analysis via Dyck Reachability 2 A. Pavlogiannis

More information

Dynamic Analysis of Inefficiently-Used Containers

Dynamic Analysis of Inefficiently-Used Containers Dynamic Analysis of Inefficiently-Used Containers Shengqian (Victor) Yang 1 Dacong Yan 1 Guoqing Xu 2 Atanas Rountev 1 1 Ohio State University 2 University of California, Irvine PRESTO: Program Analyses

More information

Part I Garbage Collection Modeling: Progress Report

Part I Garbage Collection Modeling: Progress Report Part I Garbage Collection Modeling: Progress Report Peter Libič Department of Distributed and Dependable Systems http://d3s.mff.cuni.cz CHARLES UNIVERSITY PRAGUE Faculty of Mathematics and Physics Last

More information

Efficient Runtime Tracking of Allocation Sites in Java

Efficient Runtime Tracking of Allocation Sites in Java Efficient Runtime Tracking of Allocation Sites in Java Rei Odaira, Kazunori Ogata, Kiyokuni Kawachiya, Tamiya Onodera, Toshio Nakatani IBM Research Tokyo 623-4, Shimotsuruma, Yamato-shi, Kanagawa-ken,

More information

Pointer Analysis in the Presence of Dynamic Class Loading

Pointer Analysis in the Presence of Dynamic Class Loading Pointer Analysis in the Presence of Dynamic Class Loading Martin Hirzel, Amer Diwan University of Colorado at Boulder Michael Hind IBM T.J. Watson Research Center 1 Pointer analysis motivation Code a =

More information

Screaming Fast Declarative Pointer Analysis

Screaming Fast Declarative Pointer Analysis Screaming Fast Declarative Pointer Analysis http://doop.program-analysis.org Martin Bravenboer and Yannis Smaragdakis University of Massachusetts Amherst University of Oregon NEPLS March 5, 2008 overview

More information

Large-Scale API Protocol Mining for Automated Bug Detection

Large-Scale API Protocol Mining for Automated Bug Detection Large-Scale API Protocol Mining for Automated Bug Detection Michael Pradel Department of Computer Science ETH Zurich 1 Motivation LinkedList pinconnections =...; Iterator i = pinconnections.iterator();

More information

Toward Efficient Strong Memory Model Support for the Java Platform via Hybrid Synchronization

Toward Efficient Strong Memory Model Support for the Java Platform via Hybrid Synchronization Toward Efficient Strong Memory Model Support for the Java Platform via Hybrid Synchronization Aritra Sengupta Man Cao Michael D. Bond Milind Kulkarni Ohio State University Purdue University {sengupta,caoma,mikebond}@cse.ohio-state.edu

More information

Acknowledgements These slides are based on Kathryn McKinley s slides on garbage collection as well as E Christopher Lewis s slides

Acknowledgements These slides are based on Kathryn McKinley s slides on garbage collection as well as E Christopher Lewis s slides Garbage Collection Last time Compiling Object-Oriented Languages Today Motivation behind garbage collection Garbage collection basics Garbage collection performance Specific example of using GC in C++

More information

Leveraging Test Generation and Specification Mining for Automated Bug Detection without False Positives

Leveraging Test Generation and Specification Mining for Automated Bug Detection without False Positives Leveraging Test Generation and Specification Mining for Automated Bug Detection without False Positives Michael Pradel and Thomas R. Gross Department of Computer Science ETH Zurich 1 Motivation API usage

More information

Complex, concurrent software. Precision (no false positives) Find real bugs in real executions

Complex, concurrent software. Precision (no false positives) Find real bugs in real executions Harry Xu May 2012 Complex, concurrent software Precision (no false positives) Find real bugs in real executions Need to modify JVM (e.g., object layout, GC, or ISA-level code) Need to demonstrate realism

More information

Detecting and Fixing Memory-Related Performance Problems in Managed Languages

Detecting and Fixing Memory-Related Performance Problems in Managed Languages Detecting and Fixing -Related Performance Problems in Managed Languages Lu Fang Committee: Prof. Guoqing Xu (Chair), Prof. Alex Nicolau, Prof. Brian Demsky University of California, Irvine May 26, 2017,

More information

Tail Latency: Beyond Queuing Theory. Kathryn S McKinley Xi Yang, Stephen M Blackburn, Sameh Elnikety, Yuxiong He, Ricardo Bianchini

Tail Latency: Beyond Queuing Theory. Kathryn S McKinley Xi Yang, Stephen M Blackburn, Sameh Elnikety, Yuxiong He, Ricardo Bianchini Tail Latency: Beyond Queuing Theory Kathryn S McKinley Xi Yang, Stephen M Blackburn, Sameh Elnikety, Yuxiong He, Ricardo Bianchini Microsoft s Quincy datacenter 3 Servers in US datacenters 2 Servers in

More information

CGO:U:Auto-tuning the HotSpot JVM

CGO:U:Auto-tuning the HotSpot JVM CGO:U:Auto-tuning the HotSpot JVM Milinda Fernando, Tharindu Rusira, Chalitha Perera, Chamara Philips Department of Computer Science and Engineering University of Moratuwa Sri Lanka {milinda.10, tharindurusira.10,

More information

Leak Pruning. 1. Introduction. Abstract. Department of Computer Sciences The University of Texas at Austin

Leak Pruning. 1. Introduction. Abstract. Department of Computer Sciences The University of Texas at Austin Leak Pruning Michael D. Bond Kathryn S. McKinley Department of Computer Sciences The University of Texas at Austin {mikebond,mckinley}@cs.utexas.edu Abstract Managed languages improve programmer productivity

More information

Lightweight Data Race Detection for Production Runs

Lightweight Data Race Detection for Production Runs Lightweight Data Race Detection for Production Runs Ohio State CSE Technical Report #OSU-CISRC-1/15-TR01; last updated August 2016 Swarnendu Biswas Ohio State University biswass@cse.ohio-state.edu Michael

More information

Dynamic Vertical Memory Scalability for OpenJDK Cloud Applications

Dynamic Vertical Memory Scalability for OpenJDK Cloud Applications Dynamic Vertical Memory Scalability for OpenJDK Cloud Applications Rodrigo Bruno, Paulo Ferreira: INESC-ID / Instituto Superior Técnico, University of Lisbon Ruslan Synytsky, Tetiana Fydorenchyk: Jelastic

More information

Replicating Real-Time Garbage Collector

Replicating Real-Time Garbage Collector Replicating Real-Time Garbage Collector Tomas Kalibera Purdue University, West Lafayette, IN 47907, USA; Charles University, Prague, 147 00, Czech Republic SUMMARY Real-time Java is becoming a viable platform

More information

MicroPhase: An Approach to Proactively Invoking Garbage Collection for Improved Performance

MicroPhase: An Approach to Proactively Invoking Garbage Collection for Improved Performance MicroPhase: An Approach to Proactively Invoking Garbage Collection for Improved Performance Feng Xian, Witawas Srisa-an, and Hong Jiang Department of Computer Science & Engineering University of Nebraska-Lincoln

More information

Correcting the Dynamic Call Graph Using Control Flow Constraints

Correcting the Dynamic Call Graph Using Control Flow Constraints Correting the Dynmi Cll Grph Using Control Flow Constrints Byeongheol Lee Kevin Resnik Mihel Bond Kthryn MKinley 1 Appered in CC2007 Motivtion Complexity of lrge objet oriented progrms Deompose the progrm

More information

Using Managed Runtime Systems to Tolerate Holes in Wearable Memories

Using Managed Runtime Systems to Tolerate Holes in Wearable Memories Using Managed Runtime Systems to Tolerate Holes in Wearable Memories Tiejun Gao Karin Strauss Stephen M. Blackburn Kathryn S. McKinley Doug Burger James Larus Australian National University {tiejun.gao,steve.blackburn}@anu.edu.au

More information

Jazz: A Tool for Demand-Driven Structural Testing

Jazz: A Tool for Demand-Driven Structural Testing Jazz: A Tool for Demand-Driven Structural Testing J. Misurda, J. A. Clause, J. L. Reed, P. Gandra, B. R. Childers, and M. L. Soffa Department of Computer Science University of Pittsburgh Pittsburgh, Pennsylvania

More information

On The Limits of Modeling Generational Garbage Collector Performance

On The Limits of Modeling Generational Garbage Collector Performance On The Limits of Modeling Generational Garbage Collector Performance Peter Libič Lubomír Bulej Vojtěch Horký Petr Tůma Department of Distributed and Dependable Systems Faculty of Mathematics and Physics,

More information

The Potentials and Challenges of Trace Compilation:

The Potentials and Challenges of Trace Compilation: Peng Wu IBM Research January 26, 2011 The Potentials and Challenges of Trace Compilation: Lessons learned from building a trace-jit on top of J9 JVM (Joint work with Hiroshige Hayashizaki and Hiroshi Inoue,

More information

Myths and Realities: The Performance Impact of Garbage Collection

Myths and Realities: The Performance Impact of Garbage Collection Myths and Realities: The Performance Impact of Garbage Collection Tapasya Patki February 17, 2011 1 Motivation Automatic memory management has numerous software engineering benefits from the developer

More information

Memory Monitoring in a Multi-tenant OSGi Execution Environment

Memory Monitoring in a Multi-tenant OSGi Execution Environment Memory Monitoring in a Multi-tenant OSGi Execution Environment Koutheir Attouchi, Gaël Thomas, André Bottaro, Gilles Muller To cite this version: Koutheir Attouchi, Gaël Thomas, André Bottaro, Gilles Muller.

More information

Kazunori Ogata, Dai Mikurube, Kiyokuni Kawachiya and Tamiya Onodera

Kazunori Ogata, Dai Mikurube, Kiyokuni Kawachiya and Tamiya Onodera RT0854 Computer Science 13 pages Research Report April 27, 2009 A Study of Java s non-java Memory Kazunori Ogata, Dai Mikurube, Kiyokuni Kawachiya and Tamiya Onodera IBM Research, Tokyo Research Laboratory

More information

Intelligent Compilation

Intelligent Compilation Intelligent Compilation John Cavazos Department of Computer and Information Sciences University of Delaware Autotuning and Compilers Proposition: Autotuning is a component of an Intelligent Compiler. Code

More information

The RV System Tutorial

The RV System Tutorial The RV System Tutorial Patrick Meredith and Grigore Rosu joint work with Feng Chen, Dongyun Jin, Dennis Griffith, Michael Ilseman Runtime Verification, Inc. University of Illinois The RV System!"#$%&'(%)!"#*)+,'-(!9&('/+

More information

Cachetor: Detecting Cacheable Data to Remove Bloat

Cachetor: Detecting Cacheable Data to Remove Bloat Cachetor: Detecting Cacheable Data to Remove Bloat Khanh Nguyen and Guoqing Xu University of California, Irvine, CA, USA {khanhtn, guoqingx}@ics.uci.edu ABSTRACT Modern object-oriented software commonly

More information

Performance Impact of Multithreaded Java Server Applications

Performance Impact of Multithreaded Java Server Applications Performance Impact of Multithreaded Java Server Applications Yue Luo, Lizy K. John Laboratory of Computer Architecture ECE Department University of Texas at Austin 1/2/01 1 Outline Motivation VolanoMark

More information

Copyright by Jennifer Bedke Sartor 2010

Copyright by Jennifer Bedke Sartor 2010 Copyright by Jennifer Bedke Sartor 2010 The Dissertation Committee for Jennifer Bedke Sartor certifies that this is the approved version of the following dissertation: Exploiting Language Abstraction to

More information

Lecture 10. Dynamic Analysis

Lecture 10. Dynamic Analysis Lecture 10. Dynamic Analysis Wei Le Thank Xiangyu Zhang, Michael Ernst, Tom Ball for Some of the Slides Iowa State University 2014.11 Outline What is dynamic analysis? Instrumentation Analysis Representing

More information

JIT Compilation Policy for Modern Machines

JIT Compilation Policy for Modern Machines JIT Compilation Policy for Modern Machines Prasad A. Kulkarni Department of Electrical Engineering and Computer Science, University of Kansas prasadk@ku.edu Abstract Dynamic or Just-in-Time (JIT) compilation

More information

Deriving Code Coverage Information from Profiling Data Recorded for a Trace-based Just-in-time Compiler

Deriving Code Coverage Information from Profiling Data Recorded for a Trace-based Just-in-time Compiler Deriving Code Coverage Information from Profiling Data Recorded for a Trace-based Just-in-time Compiler Christian Häubl Institute for System Software Johannes Kepler University Linz Austria haeubl@ssw.jku.at

More information

Finding the Limit: Examining the Potential and Complexity of Compilation Scheduling for JIT-Based Runtime Systems

Finding the Limit: Examining the Potential and Complexity of Compilation Scheduling for JIT-Based Runtime Systems Finding the Limit: Examining the Potential and Complexity of Compilation Scheduling for JIT-Based Runtime Systems Yufei Ding, Mingzhou Zhou, Zhijia Zhao, Sarah Eisenstat, Xipeng Shen Computer Science Department,

More information

PACER: Proportional Detection of Data Races

PACER: Proportional Detection of Data Races PACER: Proportional Detection of Data Races Michael D. Bond Katherine E. Coons Kathryn S. McKinley Department of Computer Science, The University of Texas at Austin {mikebond,coonske,mckinley}@cs.utexas.edu

More information

DELTAPATH: PRECISE AND SCALABLE CALLING CONTEXT ENCODING

DELTAPATH: PRECISE AND SCALABLE CALLING CONTEXT ENCODING DELTAPATH: PRECISE AND SCALABLE CALLING CONTEXT ENCODING Qiang Zeng*, Junghwan Rhee, Hui Zhang, Nipun Arora, Guofei Jiang, Peng Liu* NEC Laboratories America *Penn State University www.nec-labs.com Calling

More information

Copyright by Ivan Jibaja 2015

Copyright by Ivan Jibaja 2015 Copyright by Ivan Jibaja 2015 The Dissertation Committee for Ivan Jibaja certifies that this is the approved version of the following dissertation: Exploiting Hardware Heterogeneity and Parallelism for

More information

Method-Specific Dynamic Compilation using Logistic Regression

Method-Specific Dynamic Compilation using Logistic Regression Method-Specific Dynamic Compilation using Logistic Regression John Cavazos Michael F.P. O Boyle Member of HiPEAC Institute for Computing Systems Architecture (ICSA), School of Informatics University of

More information

Lightweight Data Race Detection for Production Runs

Lightweight Data Race Detection for Production Runs Lightweight Data Race Detection for Production Runs Swarnendu Biswas University of Texas at Austin (USA) sbiswas@ices.utexas.edu Man Cao Ohio State University (USA) caoma@cse.ohio-state.edu Minjia Zhang

More information

Method-Level Phase Behavior in Java Workloads

Method-Level Phase Behavior in Java Workloads Method-Level Phase Behavior in Java Workloads Andy Georges, Dries Buytaert, Lieven Eeckhout and Koen De Bosschere Ghent University Presented by Bruno Dufour dufour@cs.rutgers.edu Rutgers University DCS

More information

Compact and Efficient Strings for Java

Compact and Efficient Strings for Java Compact and Efficient Strings for Java Christian Häubl, Christian Wimmer, Hanspeter Mössenböck Institute for System Software, Christian Doppler Laboratory for Automated Software Engineering, Johannes Kepler

More information

Software Challenges: Abstraction and Analysis

Software Challenges: Abstraction and Analysis Software Challenges: Abstraction and Analysis Martin Bravenboer University of Massachusetts Amherst University of Oregon April 2, 2009 University of Waterloo my research 1 the design and implementation

More information

Static Java Program Features for Intelligent Squash Prediction

Static Java Program Features for Intelligent Squash Prediction Static Java Program Features for Intelligent Squash Prediction Jeremy Singer,, Adam Pocock, Mikel Lujan, Gavin Brown, Nikolas Ioannou, Marcelo Cintra Thread-level Speculation... Aim to use parallel multi-core

More information

An Efficient Storeless Heap Abstraction Using SSA Form

An Efficient Storeless Heap Abstraction Using SSA Form An Efficient Storeless Heap Abstraction Using SSA Form Nomair A. Naeem Ondřej Lhoták D. R. Cheriton School of Computer Science University of Waterloo, Canada {nanaeem,olhotak}@uwaterloo.ca Abstract Precise,

More information

String Deduplication for Java-based Middleware in Virtualized Environments

String Deduplication for Java-based Middleware in Virtualized Environments String Deduplication for Java-based Middleware in Virtualized Environments Michihiro Horie, Kazunori Ogata, Kiyokuni Kawachiya, Tamiya Onodera IBM Research - Tokyo Duplicated strings found on Web Application

More information

Context-sensitive points-to analysis: is it worth it?

Context-sensitive points-to analysis: is it worth it? McGill University School of Computer Science Sable Research Group Context-sensitive points-to analysis: is it worth it? Sable Technical Report No. 2005-2 Ondřej Lhoták Laurie Hendren October 21, 2005 w

More information

Boosting the Priority of Garbage: Scheduling Collection on Heterogeneous Multicore Processors

Boosting the Priority of Garbage: Scheduling Collection on Heterogeneous Multicore Processors Boosting the Priority of Garbage: Scheduling Collection on Heterogeneous Multicore Processors SHOAIB AKRAM, Ghent University JENNIFER B. SARTOR, Ghent University and Vrije Universiteit Brussel KENZO VAN

More information

Using Prefetching to Improve Reference-Counting Garbage Collectors

Using Prefetching to Improve Reference-Counting Garbage Collectors Using Prefetching to Improve Reference-Counting Garbage Collectors Harel Paz 1 and Erez Petrank 2 1 IBM Haifa Research Laboratory, Mount Carmel, Haifa 31905, ISRAEL. 2 Microsoft Research, One Microsoft

More information

Java Heap Resizing From Hacked-up Heuristics to Mathematical Models. Jeremy Singer and David R. White

Java Heap Resizing From Hacked-up Heuristics to Mathematical Models. Jeremy Singer and David R. White Java Heap Resizing From Hacked-up Heuristics to Mathematical Models Jeremy Singer and David R. White Outline Background Microeconomic Theory Heap Sizing as a Control Problem Summary Outline Background

More information

Using Prefetching to Improve Reference-Counting Garbage Collectors

Using Prefetching to Improve Reference-Counting Garbage Collectors Using Prefetching to Improve Reference-Counting Garbage Collectors Harel Paz 1, and Erez Petrank 2, 1 IBM Haifa Research Laboratory, Mount Carmel, Haifa 31905, Israel paz@il.ibm.com 2 Microsoft Research,

More information

Jinn: Synthesizing a Dynamic Bug Detector for Foreign Language Interfaces

Jinn: Synthesizing a Dynamic Bug Detector for Foreign Language Interfaces UT Austin Computer Science Technical Report TR-09-39. Under double-blind submission to PLDI 2010. Jinn: Synthesizing a Dynamic Bug Detector for Foreign Language Interfaces Byeongcheol Lee University of

More information

Our Experiences with Optimizations in Sun s Java Just-In-Time Compilers

Our Experiences with Optimizations in Sun s Java Just-In-Time Compilers Journal of Universal Computer Science, vol. 12, no. 7 (2006), 788-810 submitted: 28/1/06, accepted: 6/4/06, appeared: 28/7/06 J.UCS Our Experiences with Optimizations in Sun s Java Just-In-Time Compilers

More information

Scheduling Hard Real-time Garbage Collection

Scheduling Hard Real-time Garbage Collection Scheduling Hard Real-time Garbage Collection Tomas Kalibera, Filip Pizlo, Antony L. Hosking, Jan Vitek Purdue University Abstract Managed languages such as Java and C# are increasingly being considered

More information

Valor: Efficient, Software-Only Region Conflict Exceptions

Valor: Efficient, Software-Only Region Conflict Exceptions Valor: Efficient, Software-Only Region Conflict Exceptions Swarnendu Biswas PhD Student, Ohio State University biswass@cse.ohio-state.edu Abstract Data races complicate programming language semantics,

More information

A Quantitative Evaluation of the Contribution of Native Code to Java Workloads

A Quantitative Evaluation of the Contribution of Native Code to Java Workloads A Quantitative Evaluation of the Contribution of Native Code to Java Workloads Walter Binder University of Lugano Switzerland walter.binder@unisi.ch Jarle Hulaas, Philippe Moret EPFL Switzerland {jarle.hulaas,philippe.moret}@epfl.ch

More information

Augmenting Static Source Views in IDEs with Dynamic Metrics

Augmenting Static Source Views in IDEs with Dynamic Metrics Augmenting Static Source Views in IDEs with Dynamic Metrics David Röthlisberger 1, Marcel Härry 1, Alex Villazón 2, Danilo Ansaloni 2, Walter Binder 2, Oscar Nierstrasz 1, Philippe Moret 2 1 Software Composition

More information

Continuous Path and Edge Profiling

Continuous Path and Edge Profiling Continuous Path and Edge Profiling Michael D. Bond Kathryn S. McKinley Dept. of Computer Sciences, University of Texas at Austin E-mail: {mikebond,mckinley}@cs.utexas.edu Abstract Microarchitectures increasingly

More information

Automated Documentation Inference to Explain Failed Tests

Automated Documentation Inference to Explain Failed Tests Automated Documentation Inference to Explain Failed Tests Sai Zhang University of Washington Joint work with: Cheng Zhang, Michael D. Ernst A failed test reveals a potential bug Before bug-fixing, programmers

More information

Implementing Higher-Level Languages. Quick tour of programming language implementation techniques. From the Java level to the C level.

Implementing Higher-Level Languages. Quick tour of programming language implementation techniques. From the Java level to the C level. Implementing Higher-Level Languages Quick tour of programming language implementation techniques. From the Java level to the C level. Ahead-of-time compiler compile time C source code C compiler x86 assembly

More information