Compiler and optimization level recognition using graph neural networks - Archive ouverte HAL Accéder directement au contenu
Communication Dans Un Congrès Année : 2021

Compiler and optimization level recognition using graph neural networks

Sébastien Bardin
Tristan Benoit
Jean-Yves Marion

Résumé

We consider the problem of recovering the compiling chain used to generate a given bare binary code. We present a first attempt to devise a Graph Neural Network framework to solve this problem, in order to take into account the shallow semantics provided by the binary code's structured control flow graph (CFG). We introduce a Graph Neural Network, called Site Neural Network (SNN), dedicated to this problem. Feature extraction is simplified by forgetting almost everything in a CFG except transfer control instructions. While at an early stage, our experiments show that our method already recovers the compiler and the optimization level provenance with very high accuracy. We believe these are promising results that may offer new, more robust leads for compiling tool chain identification.
Fichier principal
Vignette du fichier
Workshop_MLBA_MLPA_2020.pdf (422.22 Ko) Télécharger le fichier
Origine : Fichiers produits par l'(les) auteur(s)

Dates et versions

hal-03270335 , version 1 (24-06-2021)

Identifiants

  • HAL Id : hal-03270335 , version 1

Citer

Sébastien Bardin, Tristan Benoit, Jean-Yves Marion. Compiler and optimization level recognition using graph neural networks. MLPA 2020 - Machine Learning for Program Analysis, Jan 2021, Yokohama / Virtual, Japan. ⟨hal-03270335⟩
197 Consultations
505 Téléchargements

Partager

Gmail Facebook X LinkedIn More