Computing stack maps with interfaces - Inria - Institut national de recherche en sciences et technologies du numérique Access content directly
Conference Papers Year : 2008

Computing stack maps with interfaces

Thomas Jensen
  • Function : Author
  • PersonId : 831246
Tiphaine Turpin
  • Function : Author
  • PersonId : 841971

Abstract

Lightweight bytecode verification uses stack maps to annotate Java bytecode programs with type information in order to reduce the verification to type checking. This paper describes an improved bytecode analyser together with algorithms for optimizing the stack maps generated. The analyser is simplified in its treatment of base values (keeping only the necessary information to ensure memory safety) and enriched in its representation of interface types, using the Dedekind-MacNeille completion technique. The computed interface information allows to remove the dynamic checks at interface method invocations. We prove the memory safety property guaranteed by the bytecode verifier using an operational semantics whose distinguishing feature is the use of un-tagged 32-bit values. For bytecode typable without sets of types we show how to prune the fix-point to obtain a stack map that can be checked without computing with sets of interfaces i.e., lightweight verification is not made more complex or costly. Experiments on three substantial test suites show that stack maps can be computed and correctly pruned by an optimized (but incomplete) pruning algorithm.
Fichier principal
Vignette du fichier
BessonJensenTurpinEcoop2008.pdf (277.21 Ko) Télécharger le fichier
Origin : Files produced by the author(s)
Loading...

Dates and versions

inria-00332526 , version 1 (21-10-2008)

Identifiers

  • HAL Id : inria-00332526 , version 1

Cite

Frédéric Besson, Thomas Jensen, Tiphaine Turpin. Computing stack maps with interfaces. ECOOP'08, 2008, Paphos, Cyprus. ⟨inria-00332526⟩
106 View
112 Download

Share

Gmail Facebook X LinkedIn More