Readings

Introducing interpreter implementation

Lua: an interpreter with a small memory footprint, often used for scripting games. Interprets instructions in a loop.

• Background on the Lua language: "The Design and Implementation of a Language for Extending Applications" by Figueiredo, Ierusalimschy and Filho. From Proceedings of XXI Brazilian Seminar on Software and Hardware (1994) pages 273-283. .ps.gz (skim): Gives language concepts and an introduction to the bytecodes.
• "The Virtual Machine of Lua 5.0" by Roberto Ierusalimschy, .pdf (skim): Slides from a presentation on 2003 of the changes between the Lua 4.0.1 virtual machine and the Lua 5.0 virtual machine

  We will read through parts of the following in class:

  Lua 4.0.1 stack machine code interpreter.
  interpreter:	http://www.lua.org/source/4.0.1/src_lvm.c.html
  machine state:	http://www.lua.org/source/4.0.1/src_lstate.h.html
  instruction formats:	http://www.lua.org/source/4.0.1/src_lopcodes.h.html
  machine limits:	http://www.lua.org/source/4.0.1/src_llimits.h.html
  Lua 5.0: register machine code intperpreter.
  interpreter:	http://www.lua.org/source/5.0/src_lvm.c.html
  machine state:	http://www.lua.org/source/5.0/src_lstate.h.html
  instruction formats:	http://www.lua.org/source/5.0/src_lopcodes.h.html
  machine limits:	http://www.lua.org/source/5.0/src_llimits.h.html

• Question to consider in class: The cost of decoding a register-machine instruction is probably greater than that of decoding a stack-machine instruction. Are there optimizations we could make to their stack-machine instruction format that would make it faster? Are there optimizations we could make to their register-machine instruction format?
• Classic papers on threaded code:
  • James R. Bell "Threaded code" CACM 1973
  • Robert B. K. Dewar "Indirect threaded code" CACM 1975
• In class: examples of of various threaded code styles from A. Ertl's thesis. "Implementation of Stack-Based Languages on Register Machines" (not on reading list.)
• An interpreter that can loop or use threaded code depending on setting in .h file: O'Caml interpreter documentation .
  (The original paper on design of this interpreter is "The Zinc Experiment: an Economical Implementation of the ML Language" INRIA tech report #117, Feb 1990 This runs to about 90 pages, describing the design of a simpified version of O'Caml. It also derives the machine from formal abstract machines for functional programming: the notation is a bit thick on pp23 - pp34 and only describes processing of functions by the machine -- not on reading list.)

  We will go over the following code from the O'Caml 3.07 distribution

  ---

  Compiling C for a stack-based VM

  There are some nice slides by Reinhart Wilhelm:
  Click on "Presentations"
  Click on "Slides of the Compiler Design Course, Summer 2001"
  There are a few introductory slides in "Abstract Machines" (under heading "Chapter 2 (revised)"
  Then proceed to "The Translation of C"

  The same page has information about compiling Pascal for the p-code machine (One of the earliest stack-based VMs) under the heading "Chapter 2".

  I don't recommend necessarily reading this material. I don't think that I am going to cover it.

  Introducing virtual machines

  J. E, Smith "An Overview of Virtual Machine Architectures"
  A good taxonomy of virtual machines.

  Optimizing interpreters

  • Todd A. Proebsting "Optimizing an ANSI C Interpreter with Superoperators"
    One way of combining operations in a VM
  • M.A. Ertl et. al. "vmgen -- A Generator of Efficient Virtual Machine Interpreters"
    Several optimization techniques, and discussion performance on various hardware.
  • Ian Piumarta and Fabio Riccardi "Optimizing Direct Threaded Code by Selective Inlining"
    Creating one VM instruction per basic block
  • Problem (suggested by C. Wolff): Proebsting's paper would benefit from a block diagram to indicate which executables generate which files. Create such a diagram What more would you need to see in the paper to make this problem easier.

  Java and CLR

  • Tim Lindholm and Frank Yellin The JavaTM Virtual Machine Specification Second Edition
    Read Chapter 3 "The Structure of the Java Virtual Machine" Browse the rest of the book as you have time, or consider it as a resource when needed.
  • Erik Miejer & John Gough "Technical Overview of the Common Language Runtime" or at CiteSeer
  • John Gough "Stacking them up: a Comparison of Virtual Machines" or at CiteSeer
  • Question: Can you extend Gough's comparison. How is the JVM geared towards interpretation and the CLR geared towards compilation? How is the JVM Java-specific. Can you think of language features that the CLR would have trouble supporting? According to Citeseer extracts, "A Portable Virtual Machine for Program Debugging and Directing" http://citeseer.nj.nec.com/573315.html may have some comments -- I haven't read it yet.
  • Etienne M. Gagnon, Laurie J. Hendren "SableVM: A Research Framework for the Efficient Execution of Java Bytecode" or at CiteSeer
  • Question: What Java-specific features are supported by Sable? Would these be useful, in present or adapted form for other languages. (Again: How Java-specific is JVM? We now have a writeup on an implementation.)

  Problem: cruising CiteSeer. Look up "Dynamically managing processor temperature and power" on Google follow the first link to CiteSeer. By working forward and backward in bibliographic citations, see if you can find some interesting papers on using program monitoring techniques to control power consumption. Deliverable: a couple of abstracts of papers.

  Same process, same deliverable, but start with title "A Framework for Dynamic Energy Efficiency and Temperature Management"

  Staged compilation and JIT compilation

  Here are some papers on code generation for JITS

  • Dawson Engler "VCODE: A Retargetable, Extensible, Very Fast Dynamic Code Generation System" (1996)
    Used for explic dynamic code generation for the 'C (tickC) staged compiler rather than automatic code generation. Extreme tradeoff: low compilation time for not great quality code.
  • I am trying to get the author of the original paper on the Kaffe JIT to send me a copy, since I can't find a copy online. I can't find author's current address...
  • Ali-Reza Adl-Tabatabai, Michael Cierniak, Guei-Yuan Lueh, Vishesh M. Parikh, James M. Stichnoth "Fast, Effective Code Generation in a Just-In-Time Java Compiler" (1998)
    The Intel JIT
  • The old and newer implementations of the CACAO JIT: how did they change?
    • Andreas Krall, Reinhard Grafl "CACAO A 64 bit JavaVM Just-in-Time Compiler" (1997)
    • Andreas Krall "Efficient JavaVM Just-in-Time Compilation" (1998)
  • Michael Chen and Kunle Olukotun "Targeting Dynamic Compilation for Embedded Environments" (2002)
  • a more recent data point
    T. Suganuma, T. Ogasawara, M. Takeuchi, T. Yasue, M. Kawahito, K. Ishizaki, H. Komatsu, and T. Nakatani "Overview of the IBM Java Just-in-Time compiler" IBM Systems Journal, 39(1):175--193, 2000. 2, 1.1.1

  I suggest skimming these, and compare:
  (1) What optimizations do they make over just stringing instructions together?
  (2) How long (in passes over source code / IR) does it take to make the optimizations?
  (3) What compromises do they (say they) make in terms of implementing standard algorithms?
  (4) Do they have breakdowns of compile time by optimization?
  (5) Do they have compile time / run time breakdowns for benchmarks?
  (6) How do the implementations of shadow stacks differ, if using shadow stacks?

  This is a lot of reading. Please skim and make comparisons.

  Adaptive (dynamic) optimization

  • Matthew Arnold, Stephen Fink, David Grove, Michael Hind, Peter F. Swe "Adaptive Optimization in the Jalapeño JVM" (2000)
  • Matthew Arnold, Michael Hind, Barbara G. Ryder "Online Feedback-Directed Optimization of Java" (OOPSLA 2002)

    The Gory Details! Not on reading list:

    • Matthew Arnold "Online Profiling And Feedback-Directed Optimization Of Java" PhD. thesis, circa 100 pages.
    • Matthew Arnold "Online Instrumentation and Feedback-Directed Optimization of Java" (2002) Tech Report.
  • Vasanth Bala, Evelyn Duesterwald, Sanjeev Banerjia "Dynamo: A Transparent Dynamic Optimization System" (2001)

    The Gory Details! Not on reading list:
    there is also the short tech report from 1999 (17 pages): http://www.hpl.hp.com/techreports/1999/HPL-1999-77.html
    and the long tech report from 1999 (102 pages, and I haven't read it!) http://www.hpl.hp.com/techreports/1999/HPL-1999-78.html

  • This one removed from reading list: too much else to read for this week...
    Derek Bruening, Evelyn Duesterwald "Design and Implementation of a Dynamic Optimization Framework for Windows" 4th ACM Workshop on Feedback-Directed and Dynamic Optimization (FDDO-4), December, 2001.
  • Derek Bruening, Timothy Garnett, Saman Amarasinghe "An Infrastructure for Adaptive Dynamic Optimization" (2003)
  • Giuseppe Desoli, Nikolay Mateev, Evelyn Duesterwald, Paolo Faraboschi, and Joseph A. Fisher "DELI: A New Run-Time Control Point"
  • This one regretfully removed from reading list: too much else to read for this week...
    S. Subramanya Sastry, Rastislav Bodík "Program Specialization for Dynamic Optimizers" (2003)
  • This one removed from reading list: too much else to read for this week...
    White Paper The Java HotSpot Virtual Machine, v1.4.0 (Did not look at in course)

  Some things to think about:
  (1) How is focus of Jikes different from Dynamo?
  (2) How has DynamoRIO evolved from Dynamo
  (3) How is DELI related to Dynamo? To DynamoRIO?
  (4) If we started a dynamic optimization project today, what sort of infrastructure would we want, what features would we work on first?
  (5) How do Dynamo, Deli fit into Jim Smith's taxonomy of virtual machines? If they don't fit then why not?

  Profiling and machine counters

  • Tom Ball and Jim Larus "Efficient Path Profiling" (1996)
  • J. Anderson et al "Continuous Profiling: Where have all the cycles gone?"

  Dynamic translation in hardware

  • Kemal Ebciolu and Erik R. Altman "DAISY: Dynamic Compilation for 100% Architectural Compatibility"
    The project page has more info.
  • Alexander Klaiber "The Technology Behind Crusoe Processors"
  • James C. Dehnert, Brian K. Grant, John P. Banning, et. al. "The Transmeta Code Morphing Software: Using Speculation, Recovery, and Adaptive Retranslation to Address Real-Life Challenges" (2003)

  Dynamic and semi-dynamic binary translation in software

  • Anton Chernoff et. al. "FX!32 A Profile-Directed Binary Translator" MICRO'98
    A better choice for this course would have been: R. J. Hookway and M. A. Herdeg "Digital FX!32: Combining Emulation and Binary Translation" Digital Technical Journal, Jan. 1997, pp. 3-12.
  • David Ung and Chistina Cifuentes "Machine-Adaptable Dynamic Binary Translation" DYNAMO'00
  • From a brief glance, Aries looks similar to some systems that we have covered already, but lets discuss it anyway:
    Cindy Zheng and Carol Thompson "PA-RISC to IA-64: Transparent Execution, No Recompilation"

  Register allocation in JITs

  Present in class: The Chaitin graph-coloring allocator and/or variants and its complexity.

  • Byung-Sun Yang et. al. "LaTTe: A Java VM Just-in-Time Compiler with Fast and Efficient Register Allocation" (PACT 1999)
  • Massimiliano Poletto and Vivek Sarkar "Linear scan register allocation" (The Jikes register allocator.)
  • Bowen Alpern et. al. "Experiences Porting the Jikes RVM to Linux/IA32"

  Security in VMs: Software fault isolation

  Present in class: reference monitors

  • Robert Wahbe et. al. "Efficient software-based fault isolation" (1994)
  • Chris Small and Margo Seltzer "MiSFIT: A Tool For Constructing Safe Extensible C++ Systems" (1997)

  Security: Stack inspection in JVM

  • Wallach, D. and E. Felten "Understanding Java Stack Inspection" In Proceedings of the 1998 IEEE Symposium on Security and Privacy, Oakland, CA, 1998.
  • Cédric Fournet and Andrew D. Gordon "Stack Inspection: Theory and Variants" (only read the non-technical portions: sections 1 -- 3 and a bit at end.)
  • Ulfar Erlingsson, Fred B. Schneider "IRM Enforcement of Java Stack Inspection"
  • Not on reading list.
    Gary McGraw and Ed Felten Securing Java, Chapter 3, Section 6. Stack inspection

  All the papers are doing the same thing. Compare terminology. What matches up with what between papers?

  Software watermarking

  
  At Chris' request

  The following may provide background"
  • Jasvir Nagra, Clark Thomborson, Christian Collberg "A Functional Taxonomy for Software Watermarking"
    The presentation slides request that people use different names than those in the paper. Also, broader defn of software watermarking used in other papers: other papers concern themselves with invisible, robust software watermarking.
  • Christian Collberg, Clark Thomborson "Software Watermarking: Models and Dynamic Embeddings"
  • Jens Palsberg et. al. "Experience with Software Watermarking"
    This is an experimental study based on the previous paper. There is supposedly a typo in the formula for counting planted planar cubic trees in this paper: see "Tamper-proofing Software Watermarks"
  • Patrick and Radhia Cousot "An Abstract Interpretation-Based Framework for Software Watermarking"
    You can ignore the math used for proving dataflow analyses (Dragon Book Chapter 10) correct, and just think in terms of what you know of constant propagation / constant folding -- will review in class.

  Tamper-proofing and obfuscation

  • White Paper "ez platform 2.0: Technical Overview"
  • B. Horne, et. al. "Dynamic Self-Checking Techniques for Improved Tamper Resistance" Digital Rights Management Workshop, Lecture Notes in Comp. Sci. 2320, Springer, 2002.
  • (optional) D. Lie, et. al. "Architectural Support for Copy and Tamper Resistant Software" Proc. 9th Int. Conf. on Architectural Support for Prog. Languages and Systems (ASPLOS-IX), November 2000.
  Obfuscation:
  • PreEmptive Solutions, Inc. "dotfuscator: Producing Smaller, More Secure .NET Applications, Technical White Paper" Version 1.1, 2002.
  • Christian Collberg, Clark Thomborson, Douglas Low "A Taxonomy of Obfuscating Transformations"

  Virtual OS

  • R.J. Creasy "The Origin of the VM/370 Time-Sharing System" IBM J. Res. Develop., 25(5):483-490, September 1981.
  • J. Sugerman, et. al. "Virtualizing I/O Devices on VMware Workstation's Hosted Virtual Machine Monitor" Proc. 2001 USENIX Annual Technical Conf., June 2001.
  • C. Waldspurger "Memory Resource Management in VMware ESX Server" Proc. 5th UNENIX Symp. on OS Design and Implementation, December 2002. (optional)

  O'Caml interpreter:	byterun/interp.c
  macros and defines:	byterun/mlvalues.h