This repo contains experiments for DivCon. For running these experiments as is, you need to have available:

• 5 physical cores
• 50GB of memory (less will also probably work)

All scripts require Python 2, e.g.:

python -V
Python 2.7.12

A way to switch to Python 2 is by using alias:

alias python="/usr/bin/python2.7"

More ways are described here

The scripts in this code have dependencies on ROPgadget, capstone, matplotlib, numpy, and uncertainties

To install these run:

pip install capstone matplotlib ROPgadget numpy uncertainties

You may need to install tk:

sudo apt-get install python-tk

To Install DivCon, follow the instruction in DivCon.

After successfully installing DivCon, set environment variable DIVCON_PATH:

export DIVCON_PATH=/path/to/divcon

Clone the git repository with Unison extension:

git clone https://github.com/unison-code/llvm.git

Select the unsion branch:

git checkout -b release_38-unison origin/release_38-unison

Install llc:

mkdir build
cd build
cmake -G "Unix Makefiles" ..
make && sudo make install

To run the experiments in parallel write:

make run-par

NOTE: The experiments are spawned to the background and might take a while before finishing.

make aggr

NOTE: The experiments are spawned to the background and might take a while before finishing.

then:

make merge

and then :

make results

make gadgets

NOTE: The experiments are spawned to the background and might take a while before finishing.

Extract .csv files

make extract