Harry Xu May 2012 Complex, concurrent software Precision (no false - - PowerPoint PPT Presentation

▶
harry xu may 2012 complex concurrent software precision
SMART_READER_LITE
LIVE PREVIEW

Harry Xu May 2012 Complex, concurrent software Precision (no false - - PowerPoint PPT Presentation

Jan 04, 2023 390 likes •1.06k views

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 (usually performance) Otherwise use

slide-1
SLIDE 1

Harry Xu May 2012

slide-2
SLIDE 2

Complex, concurrent software Precision (no false positives)

Find real bugs in real executions

slide-3
SLIDE 3

Need to modify JVM

(e.g., object layout, GC, or ISA-level code)

Need to demonstrate realism

(usually performance)

slide-4
SLIDE 4

Otherwise use RoadRunner, BCEL, Pin, LLVM, …

slide-5
SLIDE 5

Keeping track of stuff as the program executes?

 Change application behavior (add instrumentation)  Store per-object/per-field metadata  Piggyback on GC

slide-6
SLIDE 6

Keeping track of stuff as the program executes?

 JVM written in Java?!  Change application behavior (add instrumentation)  Store per-object/per-field metadata  Piggyback on GC  Uninterruptible code

slide-7
SLIDE 7

 Guide  Research Archive  Research mailing list

Jikes RVM {

slide-8
SLIDE 8

 Guide  Research Archive  Research mailing list

Jikes RVM {

slide-9
SLIDE 9

Jikes RVM source code Dynamic compilers Boot image writer

slide-10
SLIDE 10

Jikes RVM source code Run with another JVM Dynamic compilers Boot image writer

slide-11
SLIDE 11

Jikes RVM source code Boot image (native code + initial heap space) Run with another JVM Dynamic compilers Boot image writer

slide-12
SLIDE 12

Jikes RVM source code Boot image (native code + initial heap space) Run with another JVM Dynamic compilers Boot image writer Build configurations: BaseBase BaseAdaptive FullAdaptive FastAdaptive

slide-13
SLIDE 13

Jikes RVM source code Boot image (native code + initial heap space) Run with another JVM Dynamic compilers Boot image writer Build configurations: BaseBase (prototype) BaseAdaptive (prototype-opt) FullAdaptive (development) FastAdaptive (production)

slide-14
SLIDE 14

Jikes RVM source code Boot image (native code + initial heap space) Run with another JVM Dynamic compilers Boot image writer Build configurations: BaseBase BaseAdaptive FullAdaptive FastAdaptive Testing

slide-15
SLIDE 15

Jikes RVM source code Boot image (native code + initial heap space) Run with another JVM Dynamic compilers Boot image writer Build configurations: BaseBase BaseAdaptive FullAdaptive FastAdaptive Faster builds

slide-16
SLIDE 16

Jikes RVM source code Boot image (native code + initial heap space) Run with another JVM Dynamic compilers Boot image writer Build configurations: BaseBase BaseAdaptive FullAdaptive FastAdaptive Faster runs

slide-17
SLIDE 17

Jikes RVM source code Boot image (native code + initial heap space) Run with another JVM Dynamic compilers Boot image writer Build configurations: BaseBase BaseAdaptive FullAdaptive FastAdaptive Performance

slide-18
SLIDE 18

Jikes RVM source code Boot image (native code + initial heap space) Run with another JVM Dynamic compilers Boot image writer

slide-19
SLIDE 19

Edit with Eclipse (see Guide)

Jikes RVM source code Boot image (native code + initial heap space) Run with another JVM Dynamic compilers Boot image writer

slide-20
SLIDE 20

Keeping track of stuff as the program executes?

 Change application behavior (add instrumentation)  Store per-object/per-field metadata  Piggyback on GC

slide-21
SLIDE 21

Bytecode Native code Baseline compiler

slide-22
SLIDE 22

Bytecode Native code Baseline compiler Each bytecode  several x86 instructions (BaselineCompilerImpl.java)

slide-23
SLIDE 23

Bytecode Native code Baseline compiler Each bytecode  several x86 instructions (BaselineCompilerImpl.java)

slide-24
SLIDE 24
slide-25
SLIDE 25

Bytecode Native code Baseline compiler Profiling Adaptive optimization system

slide-26
SLIDE 26

Bytecode Native code Baseline compiler Profiling Adaptive optimization system Optimizing compiler (Faster) native code

slide-27
SLIDE 27

Bytecode Native code Baseline compiler Profiling Adaptive optimization system Optimizing compiler (Faster) native code

slide-28
SLIDE 28

Bytecode Optimizing compiler (Faster) native code

slide-29
SLIDE 29

Bytecode (Faster) native code HIR LIR MIR Resembles bytecode Resembles assembly code Resembles typical compiler IR (3-address code)

slide-30
SLIDE 30

Bytecode (Faster) native code HIR LIR MIR (Even faster) native code Opt levels: 0, 1, 2

slide-31
SLIDE 31

Bytecode (Faster) native code HIR LIR MIR ExpandRuntimeServices.java Add instrumentation at reads, writes, allocation, synchronization

slide-32
SLIDE 32
slide-33
SLIDE 33
slide-34
SLIDE 34
slide-35
SLIDE 35

Keeping track of stuff as the program executes?

 Change application behavior (add instrumentation)  Store per-object/per-field metadata  Piggyback on GC

slide-36
SLIDE 36

field0 field1 field2 header Low address High address

slide-37
SLIDE 37

field0 field1 field2 type info block locking & GC Object reference

slide-38
SLIDE 38

field0 field1 field2 type info block locking & GC Object reference Array length elem0 elem1 type info block locking & GC Object reference

slide-39
SLIDE 39

field0 field1 field2 type info block locking & GC Object reference Steal bits

slide-40
SLIDE 40

field0 field1 field2 type info block locking & GC Object reference misc MiscHeader.java

slide-41
SLIDE 41

field0 field1 field2 type info block locking & GC Object reference counter

slide-42
SLIDE 42

field0 field1 field2 type info block locking & GC Object reference counter Magic! Compiles down to three x86 instructions

slide-43
SLIDE 43

field0 field1 field2 type info block locking & GC Object reference counter Gotcha: can’t actually use LSB of leftmost word 2

slide-44
SLIDE 44

field0 field1 field2 type info block locking & GC Object reference counter What’s the problem with this code? 2

slide-45
SLIDE 45

field0 field1 field2 type info block locking & GC Object reference counter 2

slide-46
SLIDE 46

field0 field1 field2 type info block locking & GC Object reference not used

slide-47
SLIDE 47

field0 field1 field2 type info block locking & GC Object reference not used Compiles down to three x86 instructions

slide-48
SLIDE 48

field0 field1 field2 type info block locking & GC Object reference misc

slide-49
SLIDE 49

field0 field1 field2 type info block locking & GC Object reference misc What if GC moves object? What if GC collects object?

slide-50
SLIDE 50

Keeping track of stuff as the program executes?

 Change application behavior (add instrumentation)  Store per-object/per-field metadata  Piggyback on GC

slide-51
SLIDE 51

field0 field1 field2 type info block locking & GC Object reference // Initially worklist populated with roots while worklist has elements Object obj = worklist.pop() foreach reference field obj.f

  • bj.f = markAndPossiblyCopy(obj.f)

worklist.push(obj.f)

slide-52
SLIDE 52

field0 field1 field2 type info block locking & GC Object reference // Initially worklist populated with roots while worklist has elements Object obj = worklist.pop() foreach reference field obj.f

  • bj.f = markAndPossiblyCopy(obj.f)

worklist.push(obj.f)

slide-53
SLIDE 53

field0 field1 field2 type info block locking & GC Object reference // Initially worklist populated with roots while worklist has elements Object obj = worklist.pop() foreach reference field obj.f

  • bj.f = markAndPossiblyCopy(obj.f)

worklist.push(obj.f)

slide-54
SLIDE 54

field0 field1 field2 type info block locking & GC Object reference misc // Initially worklist populated with roots while worklist has elements Object obj = worklist.pop() foreach reference field obj.f

  • bj.f = markAndPossiblyCopy(obj.f)

worklist.push(obj.f)

  • bj.misc = markAndPossiblyCopy(obj.f)

worklist.push(obj.misc)

slide-55
SLIDE 55

field0 field1 field2 type info block locking & GC Object reference misc // Initially worklist populated with roots while worklist has elements Object obj = worklist.pop() foreach reference field obj.f

  • bj.f = markAndPossiblyCopy(obj.f)

worklist.push(obj.f)

  • bj.misc = markAndPossiblyCopy(obj.f)

worklist.push(obj.misc)

slide-56
SLIDE 56

field0 field1 field2 type info block locking & GC Object reference misc // Initially worklist populated with roots while worklist has elements Object obj = worklist.pop() foreach reference field obj.f

  • bj.f = markAndPossiblyCopy(obj.f)

worklist.push(obj.f)

  • bj.misc = markAndPossiblyCopy(obj.f)

worklist.push(obj.misc) TraceLocal.scanObject()

slide-57
SLIDE 57

field0 field1 field2 type info block locking & GC Object reference // Initially worklist populated with roots while worklist has elements Object obj = worklist.pop() foreach reference field obj.f

  • bj.f = markAndPossiblyCopy(obj.f)

worklist.push(obj.f)

slide-58
SLIDE 58

field0 field1 field2 type info block locking & GC Object reference // Initially worklist populated with roots while worklist has elements Object obj = worklist.pop() foreach reference field obj.f

  • bj.f = markAndPossiblyCopy(obj.f)

worklist.push(obj.f) TraceLocal.processNode()

slide-59
SLIDE 59

Keeping track of stuff as the program executes?

 Change application behavior (add instrumentation)  Store per-object/per-field metadata  Piggyback on GC  Uninterruptible code

slide-60
SLIDE 60

 Normal application code can be interrupted

  • Allocation  GC
  • Synchronization & yield points  join a GC

 Some VM code shouldn’t be interrupted

  • Heap etc. in inconsistent state

 Most instrumentation can’t be interrupted

  • Reads & writes aren’t GC-safe points
slide-61
SLIDE 61

@Uninterruptible static void myMethod(Object o) { // No allocation or synchronization // No calls to interruptible methods }

slide-62
SLIDE 62

@Uninterruptible static void myMethod(Object o) { currentThread.deferGC = true; Metadata m = new Metadata(); currentThread.deferGC = false; setMiscHeader(o, offset, m); }

slide-63
SLIDE 63

Need to modify JVM internals Need to demonstrate realism

 Guide  Research Archive  Research mailing list

Jikes RVM

Overview of other tasks & components Dynamic analysis examples Help (especially for novices)

slide-64
SLIDE 64

Object layout

  • Extra bits or words in header
  • Stealing bits from references
  • Discuss magic here

Adding instrumentation

  • Baseline & optimizing compilers
  • Allocation sites; reads & writes
  • Inlining instrumentation

Garbage collection

  • Piggybacking on GC
  • New spaces

Low-level stuff

  • Uninterruptible code
  • Walking the stack

Concurrency

  • Atomic stores
  • Thread-local data