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}
}
```