Re: Dataflow and function calls

gopi <gopi.onthemove@gmail.com>
Tue, 28 Jul 2009 06:18:47 -0700 (PDT)

          From comp.compilers
Related articles
Dataflow and function calls tetra2005@googlemail.com (Yuri Gribov) (2009-07-23)
Re: Dataflow and function calls gopi.onthemove@gmail.com (gopi) (2009-07-28)
Re: Dataflow and function calls tetra2005@googlemail.com (Yuri Gribov) (2009-07-29)
Re: Dataflow and function calls kamalpr@gmail.com (kamal) (2009-08-01)
| List of all articles for this month |
From: gopi <gopi.onthemove@gmail.com>
Newsgroups: comp.compilers
Date: Tue, 28 Jul 2009 06:18:47 -0700 (PDT)
Organization: Compilers Central
References: 09-07-084
Keywords: analysis, dataflow
Posted-Date: 29 Jul 2009 08:39:32 EDT

> Now comes the question. Imagine that I want to calculate reach-defs
> via standard dataflow techniques (we use simple three-address code
> without SSA). How should I handle call instructions? Should call kill
> all definitions or keep them?


A call is handled by taking advantage of type and scope information
(local variables not impacted by call), and optionally through inter-
procedural-analysis. Where there are pointers, as a pointer (in a
language like C) can point to almost anything, optimizers make
assumptions that are controlled by command line options.


A good look at optimization and compile options (gcc will do) should
give a lot more information.


gopi


---


Gopi Kumar Bulusu
Sankhya Technologies Private Limited
http://www.sankhya.com
Mobile: +91 94408 78042


US (Voice-Mail)


(408) 556-9757


Post a followup to this message

Return to the comp.compilers page.
Search the comp.compilers archives again.