devign

Our implementation of Devign

Paper: Devign: Effective Vulnerability Identification by Learning Comprehensive Program Semantics via Graph Neural Networks

Paper Authors: Yaqin Zhou, Shangqing Liu, Jingkai Siow, Xiaoning Du, Yang Liu

---

Prerequisites

Joern (https://joern.io/docs/) Python (=>3.6)(https://www.python.org/)

Python Libraries: Pandas (>=1.0.1) (https://pandas.pydata.org/) scikit-learn (>=0.22.2) (https://scikit-learn.org/stable/) PyTorch (>=1.4.0) (https://pytorch.org/) PyTorch Geometric (>=1.4.2) (https://pytorch-geometric.readthedocs.io/en/latest/notes/installation.html) Gensim (>=3.8.1) (https://radimrehurek.com/gensim/) cpgclientlib (>=0.11.111) (https://pypi.org/project/cpgclientlib/)

---

Terminal commands to execute before executing below command: python3 main.py openssl OpenSSL_0_9_6 clean

1. sudo apt install python3-pip
2. pip install gensim==3.8.1
3. pip install torch torchvision torchaudio or for cpu pip install torch==1.10.1+cpu torchvision==0.11.2+cpu torchaudio==0.10.1+cpu -f https://download.pytorch.org/whl/cpu/torch_stable.html OR for GPU pip install torch==1.10.0+cu113 torchvision==0.11.1+cu113 torchaudio==0.10.0+cu113 -f https://download.pytorch.org/whl/cu113/torch_stable.html 3.a) to verify - python3 -c "import torch; print(torch.version.cuda)" OR python3 import torch x = torch.rand(5, 3) print(x) torch.cuda.is_available() True torch.cuda.current_device() 0 torch.cuda.device(0) <torch.cuda.device object at 0x7fc520ca9280> torch.cuda.device_count() 1 torch.cuda.get_device_name(0) 'NVIDIA GeForce RTX 3080' exit()
4. pip install --upgrade pandas
5. for cpu pip install torch-scatter torch-sparse torch-geometric -f https://data.pyg.org/whl/torch-1.10.0+cpu.html for gpu pip install torch-scatter torch-sparse torch-geometric -f https://data.pyg.org/whl/torch-1.10.0+cu113.html
6. pip install cpgclientlib 6.a) to verify - python3 from cpgclient.CpgClient import CpgClient exit()

---

Below files are also executables, please provide permissions accordingly:

1. TROVON/devign/joern/joern-cli/joern-parse
2. TROVON/devign/joern/joern-cli/joern
3. TROVON/devign/joern/joern-cli/bin/joern-parse
4. TROVON/devign/joern/joern-cli/bin/ammonite-bridge

---

Execute with:

• python3 main.py [linux_kernel/openssl/wireshark] [v3_15/OpenSSL_0_9_6/v2_0_0] [clean/noisy]

for all releases of all projects with both settings ( 1. clean data training setting, 2. realistic training setting ), sequentially -

• python3 main.py

for all releases of linux_kernel project with both settings ( 1. clean data training setting, 2. realistic training setting ), sequentially -

• python3 main.py linux_kernel

for release v3.15 of linux_kernel project with both settings ( 1. clean data training setting, 2. realistic training setting ), sequentially -

• python3 main.py linux_kernel v3_15

for release v3.15 of linux_kernel project with only realistic training setting -

• python3 main.py linux_kernel v3_15 noisy

---

Parameters used:

"learning_rate" : 1e-4

"weight_decay" : 1.3e-6

"loss_lambda" : 1.3e-6

"epochs" : 100

"patience" : 10

"batch_size" : 8

"dataset_ratio" : 0.5

"shuffle" : false