Reconstructing control flow from predicated assembly code

1Citations
Citations of this article
7Readers
Mendeley users who have this article in their library.
Get full text

Abstract

Predicated instructions are a feature more and more common in contemporary instruction set architectures. Machine instructions are only executed if an individual guard register associated with the instruction evaluates to true. This enhances execution efficiency, but comes at a price: the control flow of a program is not explicit any more. Instead instructions from the same basic block may belong to different execution paths if they are subject to disjoint guard predicates. Postpass tools processing machine code with the purpose of program analyses or optimizations require the control flow graph of the input program to be known. The effectiveness of postpass analyses and optimizations strongly depends on the precision of the control flow reconstruction. If traditional reconstruction techniques are applied for processors with predicated instructions, their precision is seriously deteriorated. In this paper a generic algorithm is presented that can precisely reconstruct control flow from predicated assembly code. The algorithm is incorporated in the PROPAN system that enables high-quality machine-dependent post-pass optimizers to be generated from a concise hardware specification. The control flow reconstruction algorithm is machine-independent, and automatically derives the required hardware-specific knowledge from the machine specification. Experimental results obtained for the Philips Tri-Media TM1000 processor show that the precision of the reconstructed control flow is significantly higher than with reconstruction algorithms that do not specifically take predicated instructions into account. © Springer-Verlag Berlin Heidelberg 2003.

Cite

CITATION STYLE

APA

Decker, B., & Kästner, D. (2003). Reconstructing control flow from predicated assembly code. Lecture Notes in Computer Science (Including Subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics), 2826, 81–100. https://doi.org/10.1007/978-3-540-39920-9_7

Register to see more suggestions

Mendeley helps you to discover research relevant for your work.

Already have an account?

Save time finding and organizing research with Mendeley

Sign up for free