File size: 4,726 Bytes
691aea7 2df63db 691aea7 2df63db 383a088 2df63db 1ef1c3c 2df63db 1ef1c3c 2df63db 1ef1c3c 2df63db 272c2ee 2df63db 1ef1c3c 2df63db 68a4758 2df63db 877f2e7 68a4758 2df63db 383a088 |
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 |
---
license: mit
tags:
- decompile
- binary
---
### 1. Introduction of LLM4Decompile
LLM4Decompile aims to decompile x86 assembly instructions into C. It is finetuned from Deepseek-Coder on 4B tokens of assembly-C pairs compiled from AnghaBench.
- **Github Repository:** [LLM4Compile](https://github.com/albertan017/LLM4Decompile)
- **Paper link:** For more details check out the [paper](https://arxiv.org/abs/2403.05286).
Note: The unified optimization (UO) model is trained without prior knowledge of the optimization levels (O0~O3), the average re-executability is arond 0.21.
### 2. Evaluation Results
| Model | Re-compilability | | | | | Re-executability | | | | |
|--------------------|:----------------:|:---------:|:---------:|:---------:|:---------:|:----------------:|-----------|-----------|-----------|:---------:|
| Optimization-level | O0 | O1 | O2 | O3 | Avg. | O0 | O1 | O2 | O3 | Avg. |
| GPT4 | 0.92 | 0.94 | 0.88 | 0.84 | 0.895 | 0.1341 | 0.1890 | 0.1524 | 0.0854 | 0.1402 |
| DeepSeek-Coder-33B | 0.0659 | 0.0866 | 0.1500 | 0.1463 | 0.1122 | 0.0000 | 0.0000 | 0.0000 | 0.0000 | 0.0000 |
| LLM4Decompile-1b | 0.8780 | 0.8732 | 0.8683 | 0.8378 | 0.8643 | 0.1573 | 0.0768 | 0.1000 | 0.0878 | 0.1055 |
| LLM4Decompile-6b | 0.8817 | 0.8951 | 0.8671 | 0.8476 | 0.8729 | 0.3000 | 0.1732 | 0.1988 | 0.1841 | 0.2140 |
| LLM4Decompile-33b | 0.8134 | 0.8195 | 0.8183 | 0.8305 | 0.8204 | 0.3049 | 0.1902 | 0.1817 | 0.1817 | 0.2146 |
### 3. How to Use
Note: For the UO model, it is trained without prior knowledge of the optimization levels (O0~O3), therefore, the prompt is slightly different.
Here give an example of how to use our model.
First compile the C code into binary, disassemble the binary into assembly instructions:
```python
import subprocess
import os
import re
digit_pattern = r'\b0x[a-fA-F0-9]+\b'# binary codes in Hexadecimal
zeros_pattern = r'^0+\s'#0s
OPT = ["O0", "O1", "O2", "O3"]
fileName = 'path/to/file'
with open(fileName+'.c','r') as f:#original file
c_func = f.read()
for opt_state in OPT:
output_file = fileName +'_' + opt_state
input_file = fileName+'.c'
compile_command = f'gcc -c -o {output_file}.o {input_file} -{opt_state} -lm'#compile the code with GCC on Linux
subprocess.run(compile_command, shell=True, check=True)
compile_command = f'objdump -d {output_file}.o > {output_file}.s'#disassemble the binary file into assembly instructions
subprocess.run(compile_command, shell=True, check=True)
input_asm = ''
with open(output_file+'.s') as f:#original file
asm= f.read()
asm = asm.split('Disassembly of section .text:')[-1].strip()
for tmp in asm.split('\n'):
tmp_asm = tmp.split('\t')[-1]#remove the binary code
tmp_asm = tmp_asm.split('#')[0].strip()#remove the comments
input_asm+=tmp_asm+'\n'
input_asm = re.sub(zeros_pattern, '', input_asm)
before = f"# This is the assembly code:\n"#prompt different for the UO model
after = "\n# What is the source code?\n"#prompt
input_asm_prompt = before+input_asm.strip()+after
with open(fileName +'_' + opt_state +'.asm','w',encoding='utf-8') as f:
f.write(input_asm_prompt)
```
Then use LLM4Decompile to translate the assembly instructions into C:
```python
from transformers import AutoTokenizer, AutoModelForCausalLM
import torch
model_path = 'arise-sustech/llm4decompile-6.7b-uo'
tokenizer = AutoTokenizer.from_pretrained(model_path)
model = AutoModelForCausalLM.from_pretrained(model_path,torch_dtype=torch.bfloat16).cuda()
with open(fileName +'_' + opt_state +'.asm','r') as f:#original file
asm_func = f.read()
inputs = tokenizer(asm_func, return_tensors="pt").to(model.device)
with torch.no_grad():
outputs = model.generate(**inputs, max_new_tokens=512)
c_func_decompile = tokenizer.decode(outputs[0][len(inputs[0]):-1])
```
### 4. License
This code repository is licensed under the MIT License.
### 5. Contact
If you have any questions, please raise an issue.
### 6. Citation
```
@misc{tan2024llm4decompile,
title={LLM4Decompile: Decompiling Binary Code with Large Language Models},
author={Hanzhuo Tan and Qi Luo and Jing Li and Yuqun Zhang},
year={2024},
eprint={2403.05286},
archivePrefix={arXiv},
primaryClass={cs.PL}
}
``` |