Name
Abstract | ||
|---|---|---|
The performance of contemporary object oriented languages depends on optimizations such as devirtualization, inlining, and specialization, and these in turn depend on precise call graph analysis. Existing call graph analyses do not take advantage of the information provided by the rich type systems of contemporary languages, in particular generic type arguments. Many existing approaches analyze Java bytecode, in which generic types have been erased. This paper shows that this discarded information is actually very useful as the context in a context-sensitive analysis, where it significantly improves precision and keeps the running time small. Specifically, we propose and evaluate call graph construction algorithms in which the contexts of a method are (i) the type arguments passed to its type parameters, and (ii) the static types of the arguments passed to its term parameters. The use of static types from the caller as context is effective because it allows more precise dispatch of call sites inside the callee. Our evaluation indicates that the average number of contexts required per method is small. We implement the analysis in the Dotty compiler for Scala, and evaluate it on programs that use the type-parametric Scala collections library and on the Dotty compiler itself. The context-sensitive analysis runs 1.4x faster than a context-insensitive one and discovers 20% more monomorphic call sites at the same time. When applied to method specialization, the imprecision in a context-insensitive call graph would require the average method to be cloned 22 times, whereas the context-sensitive call graph indicates a much more practical 1.00 to 1.50 clones per method. We applied the proposed analysis to automatically specialize generic methods. The resulting automatic transformation achieves the same performance as state-of-the-art techniques requiring manual annotations, while reducing the size of the generated bytecode by up to 5Ã. |
Year | DOI | Venue |
|---|---|---|
2016 | 10.1145/2983990.2983991 | OOPSLA |
Field | DocType | Citations |
Programming language,Scala,Object-oriented programming,Computer science,Static analysis,Parametric polymorphism,Call graph,Theoretical computer science,Compiler,Java bytecode,Bytecode | Conference | 4 |
PageRank | References | Authors |
0.48 | 13 | 4 |
Name | Order | Citations | PageRank |
|---|---|---|---|
| Dmitry Petrashko | 1 | 9 | 1.56 |
| Vlad Ureche | 2 | 145 | 7.78 |
| Ondrej Lhoták | 3 | 192 | 11.86 |
| Martin Odersky | 4 | 2261 | 170.39 |