About

DirGen is a tool implemented in the paper The Compiler Removed My Live Code! Exposing Buggy Dead Code Elimination via Directed Test Case Generation (under review).

This repository includes two parts:

Part 1: The raw results used in the evaluation

• See more details in the folder experiment-results

Part 2: Source code of the approach DirGen implemented in the paper

• With the tutorial on building the DirGen in the following instructions

Folder structure

DirGen
├── experiment-results/ # Directory to store experiment results
├── compiler_test.in # configuration of compiler options
├── compiler_test.pl # Perl script for executing the test program
├── setup.sh # Shell script to install required packages
├── run-dirgen.py # Python script to run the DirGen pipeline
├── README.md # This documentation file

System Requirements

• Python 3.6+
• Unbuntu 18.04 +

Quick Setup

To set up DirGen, users could execute the following to install the required packages

git clone https://github.com/AnonyGiit/DirGen
./setup.sh

Command-Line Options Explained

This script supports the following command-line options:

• --var-selection
  Description: Disables the variable selection optimization used in the getBinaries step.
  Default Behavior: Variable selection optimization is enabled by default.

• --marker-reduction
  Description: Disables the marker reduction optimization used in the testMain step.
  Default Behavior: Marker reduction is enabled by default.

• --gen-mode
  Description: Enable different modes for directed test case generation: se, random, or hybrid.
  Default Behavior: hybrid is used by default.

• --num-tests <int>
  Description: Sets the number of test iterations to run.
  Default Value: 100

Defaults

By default, the following options are set:

parser.set_defaults(
    var_selection=True,
    marker_reduction=True,
    gen_mode=hybrid
)

Usage

• Run the following to lanuch the tool

python run-dirgen.py

• Expected output

### Test main ###
cmd_list :  ['gcc-11.1.0 -w -std=c99 -O1', 'gcc-11.1.0 -w -std=c99 -O3']
Execution time:  1.1920928955078125e-06

Geting binaries ...  0
✅ Inserted 68 marker calls and moved only function definitions before main()
[+] OPT 1 is Applied: Do select interesting variables to be symbolized ...
number of symbolic variables :  47

========== Processing testcase_1 with Data-Flow Analysis ==========
[-] Number of global variables (before OPT) :  64
[+] reduced number of global variables (after OPT) :  17
 * Seed:      11615106362440524309
Analyzing binaries ...
before comparing set1:  {'67', '1'}
before comparing set2:  {'67', '1'}
intersect :  {'67', '1'}
marker interesting :  {'test1': set(), 'test2': set()}
No interesting markers ...
############### No.0 count_var_org = 64, count_var_reduced = 17; count_diff = 0; count_org_marker = 0; count_reduced_marker = 0; count_se = 0
Execution time:  1.3597609996795654

Geting binaries ...  1
✅ Inserted 133 marker calls and moved only function definitions before main()
[+] OPT 1 is Applied: Do select interesting variables to be symbolized ...
number of symbolic variables :  23

========== Processing testcase_1 with Data-Flow Analysis ==========
[-] Number of global variables (before OPT) :  33
[+] reduced number of global variables (after OPT) :  16
 * Seed:      11435502584511641187
Analyzing binaries ...
before comparing set1:  {'1', '132', '0'}
before comparing set2:  {'1', '132', '0'}

License

MIT License