LLFI_fault_injector_tools

Utilized regular expressions to parse LLFI output files and create a SQL database

database.py

This script initiate the database and calls the other scripts depending on the process type

run python database.py

Enter benchmark name: e.g gaussian

Enter process type: e.g a

process type

• a: processes and create database for all LLFI output
• p: processes and create database for profilling file
• e: processes and create database for error output
• l: processes and create database for LLVM IR file
• s: processes and create database for standard output folder
• pr: processes and create database for program output folder

process_error_output.py

Processes files in /bamboo_fi/err_output

Modification

Change suite_path = "/home/baba/rodinia_3.1/cuda/" in create_file_path function to your file path

May require more changes in create_file_path function if not using rodinia benchmark suite

process_llvmir.py

Process the LLVM IR file in /bamboo_fi after adding debug info to the LLVM file using llvm-dis -show-annotation LLVM IR file name

Modification

Change file_path = "/home/baba/rodinia_3.1/cuda/" +benchmark+ "/bamboo_fi/" +benchmark+ "_injection.ll.ll" in llvm_main funtion to your file path

process_prog_out.py

Compares the golden output with the program output files in /bamboo_fi/prog_output. Store comparison results as either 1's for true and 0's for false

Modification

create_file_path function requires changes to golden_outfile path and suite_path

process_std_output.py

Processes files in /bamboo_fi/std_output

Modification

Change suite_path = "/home/baba/rodinia_3.1/cuda/" in create_file_path function to your file path

profilling.py

Processes the bamboo.profile.txt file

Modification

Change file_path = "/home/baba/rodinia_3.1/cuda/" + benchmark + "/bamboo_fi/bamboo.profile.txt" in profilling_main funtion to your file path