From Data to Effects Dependence Graphs: Source-to-Source Transformations for C

Nelson Lossing (1, 2) , Pierre Guillou (2, 1) , Mehdi Amini (2, 1) , François Irigoin (2, 1)

1 Mines Paris - PSL (École nationale supérieure des mines de Paris)
2 CRI - Centre de Recherche en Informatique

Program optimizations, transformations and analyses are applied to intermediate representations, which usually do not include explicit variable declarations. This description level is fine for middle-ends and for source-to-source optimizers of simple languages. However, the C language is much more flexible: variable and type declarations can appear almost anywhere in source code, and they cannot become implicit in the output code of a C source-to-source compiler. We show that declaration statements can be handled like the other statements and with the same algorithms if new effect information is defined and handled by the compiler, such as writing the environment when a variable is declared and reading it when it is accessed. This extension has been used for several years in our PIPS framework and has remained compatible with its new developments such as offloading compilers for GPUs and coprocessors.

Mots clés: C Lan-guage; Declaration Scheduling; Source-to-Source Compiler; Data Dependence Graph

Domaines: Calcul parallèle, distribué et partagé [cs.DC]; Langage de programmation [cs.PL]

Nelson Lossing, Pierre Guillou, Mehdi Amini, François Irigoin. From Data to Effects Dependence Graphs: Source-to-Source Transformations for C. [Technical Report] MINES ParisTech. 2015. ⟨hal-01254426⟩