Download presentation
Presentation is loading. Please wait.
1
Incrementalized Pointer and Escape Analysis Martin Rinard MIT LCS
2
Context Unsafe languages (C, C++, …) Safe languages (Java, ML, Scheme, …) Big difference: garbage collection Goal: analyze program to safely allocate objects on the call stack instead of in garbage- collected heap Advantages No dangling references No memory leaks Disadvantages Collection overhead Collection pauses
3
void compute(d,e) ———— void multiplyAdd(a,b,c) ————————— void multiply(m) ———— void add(u,v) —————— void main(i,j) ——————— void evaluate(i,j) —————— void abs(r) ———— void scale(n,m) —————— Program With Allocation Sites
4
void compute(d,e) ———— void multiplyAdd(a,b,c) ————————— void multiply(m) ———— void add(u,v) —————— void main(i,j) ——————— void evaluate(i,j) —————— void abs(r) ———— void scale(n,m) —————— Program With Allocation Sites When program runs
5
void compute(d,e) ———— void multiplyAdd(a,b,c) ————————— void multiply(m) ———— void add(u,v) —————— void main(i,j) ——————— void evaluate(i,j) —————— void abs(r) ———— void scale(n,m) —————— Program With Allocation Sites When program runs
6
void compute(d,e) ———— void multiplyAdd(a,b,c) ————————— void multiply(m) ———— void add(u,v) —————— void main(i,j) ——————— void evaluate(i,j) —————— void abs(r) ———— void scale(n,m) —————— Program With Allocation Sites When program runs It allocates objects in heap
7
void compute(d,e) ———— void multiplyAdd(a,b,c) ————————— void multiply(m) ———— void add(u,v) —————— void main(i,j) ——————— void evaluate(i,j) —————— void abs(r) ———— void scale(n,m) —————— Program With Allocation Sites When program runs It allocates objects in heap
8
void compute(d,e) ———— void multiplyAdd(a,b,c) ————————— void multiply(m) ———— void add(u,v) —————— void main(i,j) ——————— void evaluate(i,j) —————— void abs(r) ———— void scale(n,m) —————— Program With Allocation Sites When program runs It allocates objects in heap
9
void compute(d,e) ———— void multiplyAdd(a,b,c) ————————— void multiply(m) ———— void add(u,v) —————— void main(i,j) ——————— void evaluate(i,j) —————— void abs(r) ———— void scale(n,m) —————— Program With Allocation Sites When program runs It allocates objects in heap
10
void compute(d,e) ———— void multiplyAdd(a,b,c) ————————— void multiply(m) ———— void add(u,v) —————— void main(i,j) ——————— void evaluate(i,j) —————— void abs(r) ———— void scale(n,m) —————— Program With Allocation Sites When objects become unreachable
11
void compute(d,e) ———— void multiplyAdd(a,b,c) ————————— void multiply(m) ———— void add(u,v) —————— void main(i,j) ——————— void evaluate(i,j) —————— void abs(r) ———— void scale(n,m) —————— Program With Allocation Sites When objects become unreachable Garbage collector (eventually) reclaims memory
12
void compute(d,e) ———— void multiplyAdd(a,b,c) ————————— void multiply(m) ———— void add(u,v) —————— void main(i,j) ——————— void evaluate(i,j) —————— void abs(r) ———— void scale(n,m) —————— Stack Allocation Correlate lifetimes of objects with lifetimes of procedures
13
void compute(d,e) ———— void multiplyAdd(a,b,c) ————————— void multiply(m) ———— void add(u,v) —————— void main(i,j) ——————— void evaluate(i,j) —————— void abs(r) ———— void scale(n,m) —————— Stack Allocation Allocate object on activation record of procedure
14
void compute(d,e) ———— void multiplyAdd(a,b,c) ————————— void multiply(m) ———— void add(u,v) —————— void main(i,j) ——————— void evaluate(i,j) —————— void abs(r) ———— void scale(n,m) —————— Stack Allocation When procedure returns
15
void compute(d,e) ———— void multiplyAdd(a,b,c) ————————— void multiply(m) ———— void add(u,v) —————— void main(i,j) ——————— void evaluate(i,j) —————— void abs(r) ———— void scale(n,m) —————— Stack Allocation When procedure returns Object automatically deallocated
16
Problem How does compiler determine if it is safe to allocate objects on stack? Classic problem in program analysis Standard approach Analyze whole program Use information to find “captured” objects Allocate captured objects on stack
17
Stack Allocation Percentage of Memory Allocated on Stack 0 20 40 60 80 100 barneswaterjlexdbraytracecompress Whole-Program Analysis
18
Normalized Execution Times 0 20 40 60 80 100 barneswaterjlexdbraytracecompress Whole-Program Analysis Reference: execution time without optimization
19
Analysis Times 0 25 50 75 100 125 150 barneswaterjlexdbraytracecompress Analysis Time (seconds) Whole-Program Analysis 223 645
20
void compute(d,e) ———— void multiplyAdd(a,b,c) ————————— void multiply(m) ———— void add(u,v) —————— void main(i,j) ——————— void evaluate(i,j) —————— void abs(r) ———— void scale(n,m) —————— Key Observation Number One: Most optimizations require only the analysis of a small part of program surrounding the object allocation site
21
void compute(d,e) ———— void multiplyAdd(a,b,c) ————————— void multiply(m) ———— void add(u,v) —————— void main(i,j) ——————— void evaluate(i,j) —————— void abs(r) ———— void scale(n,m) —————— Key Observation Number Two: Most of the optimization benefit comes from a small percentage of the allocation sites 99% of objects allocated at these two sites
22
Intuition for Better Analysis Locate important allocation sites Use demand-driven approach to analyze region surrounding site Somehow avoid sinking analysis resources into unprofitable sites void compute(d,e) ———— void multiplyAdd(a,b,c) ————————— void multiply(m) ———— void add(u,v) —————— void main(i,j) ——————— void evaluate(i,j) —————— void abs(r) ———— void scale(n,m) —————— 99% of objects allocated at these two sites
23
How we do it Develop an incremental, demand-driven analysis Use profiling to identify important allocation sites Monitor analysis results to estimate Chance of optimizing site Analysis cost of realizing optimization Use estimates to direct analysis to profitable sites Result Usually get almost all of the benefit of whole-program analysis But at a fraction of the analysis cost
24
What This Talk is About How we turned this intuition into an algorithm that usually 1)obtains almost all the benefit of the whole program analysis 2) analyzes a small fraction of program 3) consumes a small fraction of whole program analysis time
25
Basic Idea Incremental, demand-driven analysis Profiling identifies important allocation sites Monitor analysis results to estimate Chance of optimizing site Analysis cost of realizing optimization Estimates direct analysis to profitable sites
26
Structure of Talk Motivating Example Base whole program analysis (Whaley and Rinard, OOPSLA 99) Incrementalized analysis Analysis policy Experimental results
27
Example
28
Employee Database Example Traverse database, extract max salary Name Salary John Doe $45,000 Ben Bit $30,000 Jane Roe $55,000 Vector max salary = $55,000 highest paid = Jane Roe
29
Coding Max Computation (in Java) class EmployeeDatabase { Vector database = new Vector(); Employee highestPaid; void computeMax() { int max = 0; Enumeration enum = database.elements(); while (enum.hasMoreElements()) { Employee e = enum.nextElement(); if (max < e.salary()) { max = e.salary(); highestPaid = e; }
30
Coding Max Computation (in Java) class EmployeeDatabase { Vector database = new Vector(); Employee highestPaid; void computeMax() { int max = 0; Enumeration enum = database.elements(); while (enum.hasMoreElements()) { Employee e = enum.nextElement(); if (max < e.salary()) { max = e.salary(); highestPaid = e; } Would like to allocate enum object on stack, not on the heap
31
Whole Program Analysis
32
void computeMax() ———— Enumeration elements() ——— Vector elementData() ———— boolean hasMoreElements() —————— void printStatistics() ——————— Employee nextElement() —————— int salary() —————— Bottom Up and Compositional Currently analyzed procedure Currently analyzed part of the program
33
void computeMax() ———— Enumeration elements() ——— Vector elementData() ———— boolean hasMoreElements() —————— void printStatistics() ——————— Employee nextElement() —————— int salary() —————— Bottom Up and Compositional Currently analyzed procedure Currently analyzed part of the program
34
Base Analysis Basic Abstraction: Points-to Escape Graph Intraprocedural Analysis Flow sensitive abstract interpretation Produces points-to escape graph at each program point Interprocedural Analysis Bottom up and compositional Analyzes each procedure once to obtain a single parameterized analysis result Result is specialized for use at each call site
35
Points-to Graph in Example void computeMax() { int max = 0; Enumeration enum = database.elements(); while (enum.hasMoreElements()) { Employee e = enum.nextElement(); if (max < e.salary()) { max = e.salary(); highestPaid = e; } } vectorelementData [ ] this enum e database highestPaid
![Points-to Graph in Example void computeMax() { int max = 0; Enumeration enum = database.elements(); while (enum.hasMoreElements()) { Employee e = enum.nextElement(); if (max < e.salary()) { max = e.salary(); highestPaid = e; } } vectorelementData [ ] this enum e database highestPaid](http://images.slideplayer.com./27/9006944/slides/slide_35.jpg "Points-to Graph in Example void computeMax() { int max = 0; Enumeration enum = database.elements(); while (enum.hasMoreElements()) { Employee e = enum.nextElement(); if (max < e.salary()) { max = e.salary(); highestPaid = e; } } vectorelementData [ ] this enum e database highestPaid")
36
Edge Types Inside Edges: Represent references created in currently analyzed part of program Outside Edges: Represent references created outside currently analyzed part of program vectorelementData [ ] this enum e database highestPaid dashed = outside solid = inside
![Edge Types Inside Edges: Represent references created in currently analyzed part of program Outside Edges: Represent references created outside currently analyzed part of program vectorelementData [ ] this enum e database highestPaid dashed = outside solid = inside](http://images.slideplayer.com./27/9006944/slides/slide_36.jpg "Edge Types Inside Edges: Represent references created in currently analyzed part of program Outside Edges: Represent references created outside currently analyzed part of program vectorelementData [ ] this enum e database highestPaid dashed = outside solid = inside")
37
Node Types Inside Nodes: Represent objects created in currently analyzed part of program Outside Nodes: Parameter nodes – represent parameters Load nodes - represent objects accessed via pointers created outside analyzed part vectorelementData [ ] this enum e database highestPaid dashed = outside solid = inside
![Node Types Inside Nodes: Represent objects created in currently analyzed part of program Outside Nodes: Parameter nodes – represent parameters Load nodes - represent objects accessed via pointers created outside analyzed part vectorelementData [ ] this enum e database highestPaid dashed = outside solid = inside](http://images.slideplayer.com./27/9006944/slides/slide_37.jpg "Node Types Inside Nodes: Represent objects created in currently analyzed part of program Outside Nodes: Parameter nodes – represent parameters Load nodes - represent objects accessed via pointers created outside analyzed part vectorelementData [ ] this enum e database highestPaid dashed = outside solid = inside")
38
Escape Information Escaped nodes parameter nodes returned nodes nodes reachable from other escaped nodes Captured is the opposite of escaped vectorelementData [ ] this enum e database highestPaid green = escaped white = captured
![Escape Information Escaped nodes parameter nodes returned nodes nodes reachable from other escaped nodes Captured is the opposite of escaped vectorelementData [ ] this enum e database highestPaid green = escaped white = captured](http://images.slideplayer.com./27/9006944/slides/slide_38.jpg "Escape Information Escaped nodes parameter nodes returned nodes nodes reachable from other escaped nodes Captured is the opposite of escaped vectorelementData [ ] this enum e database highestPaid green = escaped white = captured")
39
Stack Allocation Optimization Examine graph from end of procedure If a node is captured in this graph Allocate corresponding objects on stack (may need to inline procedures to apply optimization) vectorelementData [ ] this enum e database highestPaid green = escaped white = captured Can allocate enum object on stack
![Stack Allocation Optimization Examine graph from end of procedure If a node is captured in this graph Allocate corresponding objects on stack (may need to inline procedures to apply optimization) vectorelementData [ ] this enum e database highestPaid green = escaped white = captured Can allocate enum object on stack](http://images.slideplayer.com./27/9006944/slides/slide_39.jpg "Stack Allocation Optimization Examine graph from end of procedure If a node is captured in this graph Allocate corresponding objects on stack (may need to inline procedures to apply optimization) vectorelementData [ ] this enum e database highestPaid green = escaped white = captured Can allocate enum object on stack")
40
Interprocedural Analysis void printStatistics(BufferedReader r) { EmployeeDatabase e = new EmployeeDatabase(r); e.computeMax(); System.out.println(“max salary = “ + e.highestPaid); }
41
Start with graph before call site void printStatistics(BufferedReader r) { EmployeeDatabase e = new EmployeeDatabase(r); e.computeMax(); System.out.println(“max salary = “ + e.highestPaid); } e graph before call site elementData [ ] database
![Start with graph before call site void printStatistics(BufferedReader r) { EmployeeDatabase e = new EmployeeDatabase(r); e.computeMax(); System.out.println( max salary = + e.highestPaid); } e graph before call site elementData [ ] database](http://images.slideplayer.com./27/9006944/slides/slide_41.jpg "Start with graph before call site void printStatistics(BufferedReader r) { EmployeeDatabase e = new EmployeeDatabase(r); e.computeMax(); System.out.println( max salary = + e.highestPaid); } e graph before call site elementData [ ] database")
42
Retrieve graph from end of callee void printStatistics(BufferedReader r) { EmployeeDatabase e = new EmployeeDatabase(r); e.computeMax(); System.out.println(“max salary = “ + e.highestPaid); } e graph before call site elementData [ ] database elementData [ ] this database highestPaid Enum object is not present because it was captured in the callee
![Retrieve graph from end of callee void printStatistics(BufferedReader r) { EmployeeDatabase e = new EmployeeDatabase(r); e.computeMax(); System.out.println( max salary = + e.highestPaid); } e graph before call site elementData [ ] database elementData [ ] this database highestPaid Enum object is not present because it was captured in the callee](http://images.slideplayer.com./27/9006944/slides/slide_42.jpg "Retrieve graph from end of callee void printStatistics(BufferedReader r) { EmployeeDatabase e = new EmployeeDatabase(r); e.computeMax(); System.out.println( max salary = + e.highestPaid); } e graph before call site elementData [ ] database elementData [ ] this database highestPaid Enum object is not present because it was captured in the callee")
43
Map formals to actuals void printStatistics(BufferedReader r) { EmployeeDatabase e = new EmployeeDatabase(r); e.computeMax(); System.out.println(“max salary = “ + e.highestPaid); } e graph before call site elementData [ ] database elementData [ ] this database highestPaid
![Map formals to actuals void printStatistics(BufferedReader r) { EmployeeDatabase e = new EmployeeDatabase(r); e.computeMax(); System.out.println( max salary = + e.highestPaid); } e graph before call site elementData [ ] database elementData [ ] this database highestPaid](http://images.slideplayer.com./27/9006944/slides/slide_43.jpg "Map formals to actuals void printStatistics(BufferedReader r) { EmployeeDatabase e = new EmployeeDatabase(r); e.computeMax(); System.out.println( max salary = + e.highestPaid); } e graph before call site elementData [ ] database elementData [ ] this database highestPaid")
44
Match corresponding inside and outside edges to complete mapping void printStatistics(BufferedReader r) { EmployeeDatabase e = new EmployeeDatabase(r); e.computeMax(); System.out.println(“max salary = “ + e.highestPaid); } e graph before call site elementData [ ] database elementData [ ] this database highestPaid
![Match corresponding inside and outside edges to complete mapping void printStatistics(BufferedReader r) { EmployeeDatabase e = new EmployeeDatabase(r); e.computeMax(); System.out.println( max salary = + e.highestPaid); } e graph before call site elementData [ ] database elementData [ ] this database highestPaid](http://images.slideplayer.com./27/9006944/slides/slide_44.jpg "Match corresponding inside and outside edges to complete mapping void printStatistics(BufferedReader r) { EmployeeDatabase e = new EmployeeDatabase(r); e.computeMax(); System.out.println( max salary = + e.highestPaid); } e graph before call site elementData [ ] database elementData [ ] this database highestPaid")
45
Match corresponding inside and outside edges to complete mapping void printStatistics(BufferedReader r) { EmployeeDatabase e = new EmployeeDatabase(r); e.computeMax(); System.out.println(“max salary = “ + e.highestPaid); } e graph before call site elementData [ ] database elementData [ ] this database highestPaid
![Match corresponding inside and outside edges to complete mapping void printStatistics(BufferedReader r) { EmployeeDatabase e = new EmployeeDatabase(r); e.computeMax(); System.out.println( max salary = + e.highestPaid); } e graph before call site elementData [ ] database elementData [ ] this database highestPaid](http://images.slideplayer.com./27/9006944/slides/slide_45.jpg "Match corresponding inside and outside edges to complete mapping void printStatistics(BufferedReader r) { EmployeeDatabase e = new EmployeeDatabase(r); e.computeMax(); System.out.println( max salary = + e.highestPaid); } e graph before call site elementData [ ] database elementData [ ] this database highestPaid")
46
Match corresponding inside and outside edges to complete mapping void printStatistics(BufferedReader r) { EmployeeDatabase e = new EmployeeDatabase(r); e.computeMax(); System.out.println(“max salary = “ + e.highestPaid); } e graph before call site elementData [ ] database elementData [ ] this database highestPaid
![Match corresponding inside and outside edges to complete mapping void printStatistics(BufferedReader r) { EmployeeDatabase e = new EmployeeDatabase(r); e.computeMax(); System.out.println( max salary = + e.highestPaid); } e graph before call site elementData [ ] database elementData [ ] this database highestPaid](http://images.slideplayer.com./27/9006944/slides/slide_46.jpg "Match corresponding inside and outside edges to complete mapping void printStatistics(BufferedReader r) { EmployeeDatabase e = new EmployeeDatabase(r); e.computeMax(); System.out.println( max salary = + e.highestPaid); } e graph before call site elementData [ ] database elementData [ ] this database highestPaid")
47
Match corresponding inside and outside edges to complete mapping void printStatistics(BufferedReader r) { EmployeeDatabase e = new EmployeeDatabase(r); e.computeMax(); System.out.println(“max salary = “ + e.highestPaid); } e graph before call site elementData [ ] database elementData [ ] this database highestPaid
![Match corresponding inside and outside edges to complete mapping void printStatistics(BufferedReader r) { EmployeeDatabase e = new EmployeeDatabase(r); e.computeMax(); System.out.println( max salary = + e.highestPaid); } e graph before call site elementData [ ] database elementData [ ] this database highestPaid](http://images.slideplayer.com./27/9006944/slides/slide_47.jpg "Match corresponding inside and outside edges to complete mapping void printStatistics(BufferedReader r) { EmployeeDatabase e = new EmployeeDatabase(r); e.computeMax(); System.out.println( max salary = + e.highestPaid); } e graph before call site elementData [ ] database elementData [ ] this database highestPaid")
48
Combine graphs to obtain new graph after call site void printStatistics(BufferedReader r) { EmployeeDatabase e = new EmployeeDatabase(r); e.computeMax(); System.out.println(“max salary = “ + e.highestPaid); } e graph before call site elementData [ ] database elementData [ ] this database highestPaid graph after call site e elementData [ ] highestPaid
![Combine graphs to obtain new graph after call site void printStatistics(BufferedReader r) { EmployeeDatabase e = new EmployeeDatabase(r); e.computeMax(); System.out.println( max salary = + e.highestPaid); } e graph before call site elementData [ ] database elementData [ ] this database highestPaid graph after call site e elementData [ ] highestPaid](http://images.slideplayer.com./27/9006944/slides/slide_48.jpg "Combine graphs to obtain new graph after call site void printStatistics(BufferedReader r) { EmployeeDatabase e = new EmployeeDatabase(r); e.computeMax(); System.out.println( max salary = + e.highestPaid); } e graph before call site elementData [ ] database elementData [ ] this database highestPaid graph after call site e elementData [ ] highestPaid")
49
Continue analysis after call site void printStatistics(BufferedReader r) { EmployeeDatabase e = new EmployeeDatabase(r); e.computeMax(); System.out.println(“max salary = “ + e.highestPaid); } e graph before call site elementData [ ] database graph after call site e elementData [ ] highestPaid
![Continue analysis after call site void printStatistics(BufferedReader r) { EmployeeDatabase e = new EmployeeDatabase(r); e.computeMax(); System.out.println( max salary = + e.highestPaid); } e graph before call site elementData [ ] database graph after call site e elementData [ ] highestPaid](http://images.slideplayer.com./27/9006944/slides/slide_49.jpg "Continue analysis after call site void printStatistics(BufferedReader r) { EmployeeDatabase e = new EmployeeDatabase(r); e.computeMax(); System.out.println( max salary = + e.highestPaid); } e graph before call site elementData [ ] database graph after call site e elementData [ ] highestPaid")
50
Whole Program Analysis void computeMax() ———— Enumeration elements() ——— Vector elementData() ———— boolean hasMoreElements() —————— void printStatistics() ——————— Employee nextElement() —————— int salary() ——————
51
Whole Program Analysis void computeMax() ———— Enumeration elements() ——— Vector elementData() ———— boolean hasMoreElements() —————— void printStatistics() ——————— Employee nextElement() —————— int salary() ——————
52
Whole Program Analysis void computeMax() ———— Enumeration elements() ——— Vector elementData() ———— boolean hasMoreElements() —————— void printStatistics() ——————— Employee nextElement() —————— int salary() ——————
53
Whole Program Analysis void computeMax() ———— Enumeration elements() ——— Vector elementData() ———— boolean hasMoreElements() —————— void printStatistics() ——————— Employee nextElement() —————— int salary() ——————
54
Whole Program Analysis void computeMax() ———— Enumeration elements() ——— Vector elementData() ———— boolean hasMoreElements() —————— void printStatistics() ——————— Employee nextElement() —————— int salary() ——————
55
Whole Program Analysis void computeMax() ———— Enumeration elements() ——— Vector elementData() ———— boolean hasMoreElements() —————— void printStatistics() ——————— Employee nextElement() —————— int salary() ——————
56
Whole Program Analysis void computeMax() ———— Enumeration elements() ——— Vector elementData() ———— boolean hasMoreElements() —————— void printStatistics() ——————— Employee nextElement() —————— int salary() ——————
57
Whole Program Analysis void computeMax() ———— Enumeration elements() ——— Vector elementData() ———— boolean hasMoreElements() —————— void printStatistics() ——————— Employee nextElement() —————— int salary() ——————
58
Incrementalized Analysis
59
Incrementalized Analysis Requirements Must be able to Analyze procedure independently of callers Whole-program analysis is compositional Already does this Skip analysis of invoked procedures But later incrementally integrate analysis results if desirable to do so
60
First Extension to Whole-Program Analysis Skip the analysis of invoked procedures Parameters are marked as escaping into skipped call site Assume analysis skips enum.nextElement() vector 1 this enum e database highestPaid Node 1 escapes into enum.nextElement()
61
First Extension Almost Works Can skip analysis of invoked procedures If allocation site is captured, great! If not, escape information tells you what procedures you should have analyzed… Should have analyzed enum.nextElement() vector 1 this enum e database highestPaid Node 1 escapes into enum.nextElement()
62
Second Extension to Base Algorithm Record enough information to undo skip and incorporate analysis into existing result Parameter mapping at call site Ordering information for call sites
63
void compute() { —————— foo(x,y); ——————— } Graph before call site Graph after call site Graph at end of procedure Graphs from Whole-Program Analysis Generated during analysis Used to apply stack allocation optimization
64
void compute() { —————— foo(x,y); ——————— } Graph before call site Graph after skipped call site Graph at end of procedure Graphs from Incrementalized Analysis Generated during analysis Used to apply stack allocation optimization
65
Incorporating Result from Skipped Call Site Naive approach: use mapping algorithm directly on graph from end of caller procedure Before incorporating result from skipped call site After incorporating result from skipped call site
66
Incorporating Result from Skipped Call Site Naive approach: use mapping algorithm directly on graph from end of caller procedure Graph from whole program analysis After incorporating result from skipped call site
67
Basic Problem and Solution Problem: additional edges in graph from end of method make result less precise Solution: augment abstraction For each call skipped call site, record Edges which were present in the graph before the call site Edges which were present after call site Use this information when incorporating results from skipped call sites
68
After Augmenting Abstraction Graph from whole program analysis After incorporating result from skipped call site
69
void computeMax() ———— Enumeration elements() ——— Vector elementData() ———— boolean hasMoreElements() —————— void printStatistics() ——————— Employee nextElement() —————— int salary() —————— Incrementalized Analysis
70
void computeMax() ———— Enumeration elements() ——— Vector elementData() ———— boolean hasMoreElements() —————— void printStatistics() ——————— Employee nextElement() —————— int salary() —————— Incrementalized Analysis Attempt to stack allocate Enumeration object from elements
71
void computeMax() ———— Enumeration elements() ——— Vector elementData() ———— boolean hasMoreElements() —————— void printStatistics() ——————— Employee nextElement() —————— int salary() —————— Incrementalized Analysis Analyze elements (intraprocedurally)
72
void computeMax() ———— Enumeration elements() ——— Vector elementData() ———— boolean hasMoreElements() —————— void printStatistics() ——————— Employee nextElement() —————— int salary() —————— Incrementalized Analysis Analyze elements (intraprocedurally)
73
void computeMax() ———— Enumeration elements() ——— Vector elementData() ———— boolean hasMoreElements() —————— void printStatistics() ——————— Employee nextElement() —————— int salary() —————— Incrementalized Analysis Analyze elements (intraprocedurally) Escapes only into the caller
74
void computeMax() ———— Enumeration elements() ——— Vector elementData() ———— boolean hasMoreElements() —————— void printStatistics() ——————— Employee nextElement() —————— int salary() —————— Incrementalized Analysis Analyze computeMax (intraprocedurally)
75
void computeMax() ———— Enumeration elements() ——— Vector elementData() ———— boolean hasMoreElements() —————— void printStatistics() ——————— Employee nextElement() —————— int salary() —————— Incrementalized Analysis Analyze computeMax (intraprocedurally)
76
void computeMax() ———— Enumeration elements() ——— Vector elementData() ———— boolean hasMoreElements() —————— void printStatistics() ——————— Employee nextElement() —————— int salary() —————— Incrementalized Analysis Analyze computeMax (intraprocedurally) Escapes to hasMoreElements nextElement
77
void computeMax() ———— Enumeration elements() ——— Vector elementData() ———— boolean hasMoreElements() —————— void printStatistics() ——————— Employee nextElement() —————— int salary() —————— Incrementalized Analysis Analyze hasMoreElements and nextElement (intraprocedurally)
78
void computeMax() ———— Enumeration elements() ——— Vector elementData() ———— boolean hasMoreElements() —————— void printStatistics() ——————— Employee nextElement() —————— int salary() —————— Incrementalized Analysis Incorporate Results
79
void computeMax() ———— Enumeration elements() ——— Vector elementData() ———— boolean hasMoreElements() —————— void printStatistics() ——————— Employee nextElement() —————— int salary() —————— Incrementalized Analysis Enumeration object Captured in computeMax
80
void computeMax() ———— Enumeration elements() ——— Vector elementData() ———— boolean hasMoreElements() —————— void printStatistics() ——————— Employee nextElement() —————— int salary() —————— Incrementalized Analysis Enumeration object Captured in computeMax Inline elements Stack allocate enumeration object
81
void computeMax() ———— Enumeration elements() ——— Vector elementData() ———— boolean hasMoreElements() —————— void printStatistics() ——————— Employee nextElement() —————— int salary() —————— Incrementalized Analysis We skipped the analysis of some procedures
82
void computeMax() ———— Enumeration elements() ——— Vector elementData() ———— boolean hasMoreElements() —————— void printStatistics() ——————— Employee nextElement() —————— int salary() —————— Incrementalized Analysis We skipped the analysis of some procedures We ignored some other procedures
83
Result We can incrementally analyze Only what is needed For whatever allocation site we want And even temporarily suspend analysis part of the way through!
84
New Issue We can incrementally analyze Only what is needed For whatever allocation site we want And even temporarily suspend analysis part of the way through! But… Lots of analysis opportunities Not all opportunities are profitable Where to invest analysis resources? How much resources to invest?
85
Analysis Policy Formulate policy as solution to an investment problem Goal Maximize optimization payoff from invested analysis resources
86
Analysis Policy Implementation For each allocation site, estimate marginal return on invested analysis resources Loop Invest a unit of analysis resources (time) in site that offers best return Expand analyzed region surrounding site When unit expires, recompute marginal returns (best site may change)
87
Marginal Return Estimate N · P(d) C · T N = Number of objects allocated at the site P(d) = Probability of capturing the site, knowing we explored a region of call depth d C = Number of skipped call sites the allocation site escapes through T = Average time needed to analyze a call site
88
Marginal Return Estimate N · P(d) C · T As invest analysis resources explore larger regions around allocation sites get more information about sites marginal return estimates improve analysis makes better investment decisions!
89
Usage Scenarios Ahead of time compiler Give algorithm an analysis budget Algorithm spends budget Takes whatever optimizations it uncovered Dynamic compiler Algorithm acquires analysis budget as a percentage of run time Periodically spends budget, delivers additional optimizations Longer program runs, more optimizations
90
Experimental Results
91
Methodology Implemented analysis in MIT Flex System Obtained several benchmarks Scientific computations: barnes, water Our lexical analyzer: jlex Spec benchmarks: db, raytrace, compress
92
Analysis Times 223645 jdk 1.2
93
Allocation Sites Analyzed
95
Distribution of Allocated Objects
96
Analysis Time Payoff Stack allocated by incrementalized analysis Decided Stack allocated by whole-program analysis % of Objects Analysis Time (seconds) compressraytracedb Analysis Time (seconds) barneswaterjlex
97
Stack Allocation
98
Normalized Execution Times Reference: execution time without optimization
99
Experimental Summary Key Application Properties Most objects allocated at very few allocation sites Can capture objects with an incremental analysis of region surrounding allocation sites Consequence Most of the benefits of whole-program analysis Fraction of the cost of whole-program analysis
100
Other Analyses and Uses Whole-program pointer and escape analysis (OOPSLA 1999) Stack allocation Synchronization elimination Pointer and escape analysis for multithreaded programs (PLDI 1999, PPoPP 2001) Correct use of region-based allocation Elimination of dynamic region checks Synchronization elimination Memory bank disambiguation (RAW, DeepC) Foundation for other analyses Bitwidth analysis (MIT and CMU) Symbolic accessed region analysis
101
Other Analyses and Uses Symbolic analysis of accessed regions of memory blocks (PPoPP 1999, PLDI 2000) Automatic parallelization of sequential divide and conquer programs Data race freedom for parallel divide and conquer programs Absence of array bounds violations for both Bitwidth analysis
102
Future Directions Information from designers and developers Type systems for program properties Automatic parallelization and data race freedom for divide and conquer programs (CC 2001) Data race freedom for OO programs (OOPSLA 2001) Application-specific design properties Shapes of recursive data structures Object role transitions and constraints Distributed systems, component-based systems Interaction patterns Failure propagation properties Global view of system
103
Related Work Demand-driven Analyses Horwitz, Reps, Sagiv (FSE 1995) Duesterwald, Soffa, Gupta (TOPLAS 1997) Heintze, Tardieu (PLDI 2001) Key differences Integration of escape information enables incrementalized algorithms to suspend partially completed analyses Maintain accurate marginal payoff estimates Avoid overly costly analyses
104
Related Work Previous Escape Analyses Blanchet (OOPSLA 1999) Bogda, Hoelzle (OOPSLA 1999) Choi, Gupta, Serrano, Sreedhar, Midkiff (OOPSLA 1999) Whaley, Rinard (OOPSLA 1999) Ruf (PLDI 2000) Key Differences Previous algorithms analyze whole program But could incrementalize other analyses Get a range of algorithms with varying analysis time/precision tradeoffs
105
Conclusion Whole-Program Analysis Incrementalized Analysis Properties Uses escape information to incrementally analyze relevant regions of program Analysis policy driven by estimates of optimization benefits and costs Results Most of benefits of whole program analysis Fraction of the cost
Similar presentations
