decompiler & static symbolic analysis

Vijay Ganesh <vgee@india.ti.com>
7 Nov 1996 20:43:58 -0500

          From comp.compilers

Related articles
decompiler & static symbolic analysis vgee@india.ti.com (Vijay Ganesh) (1996-11-07)
Re: decompiler & static symbolic analysis pardo@cs.washington.edu (1996-11-18)
| List of all articles for this month |
From: Vijay Ganesh <vgee@india.ti.com>
Newsgroups: comp.compilers
Date: 7 Nov 1996 20:43:58 -0500
Organization: Texas Instruments
Keywords: disassemble, question, comment

Hello,


I have got a pretty interesting problem ( at least I think so). I am
trying to convert an assembly file into C file.


As a first step to conversion :- I split the assembly file into a
'basic blocks' and then generate a call graph out of it and do control
flow analysis. To generate the call graph I identify branch, jump,
and call instructions which change the sequential flow of control.


Now I also find the target instructions of these control
instructions. this is important to identify the basic blocks.


Now, the problem is that there are some instructions which jump on the
value of a processor register. that means until run time I cannot
determine the target of these instructions.


Is there some means (in say data flow analysis) wherein I can
determine with some accuracy the probable lines of the assembly code
to which the jump may occur, statically. (The assumption is that all
the code is in a single file).


Can static program slicing be used as a means of finding the finite
set of values that the accumulator might take while executing the
branch on accumulator value instruction?


Also can anybody give me info/pointer to material/people/web
sites/papers for 'static symbolic analysis' and 'symbolic execution'.


thanks &
warm regards,


vijay ganesh.
[As you're probably discovering, decompiling assembler into useful C is
a very hard problem. -John]
--


Post a followup to this message

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