---
license: gpl-2.0
task_categories:
- text-generation
- text-classification
language:
- en
tags:
- linux
- kernel
- api-migration
- code-analysis
- software-engineering
size_categories:
- 1K<n<10K
---

# Linux Kernel API Migrations Dataset

This dataset contains API migration information extracted from the Linux kernel source code between versions **v6.8** and **v6.9**.

## Dataset Summary

- **Total Migrations**: 4,311
- **Commits Analyzed**: 15,678
- **Focus Paths**: all
- **Format**: Parquet

## Dataset Structure

### Data Fields

- **type** (object): 4311 non-null values
- **api_name** (object): 4311 non-null values
- **file** (object): 4311 non-null values
- **commit** (object): 4311 non-null values
- **old_signature** (object): 4118 non-null values
- **new_signature** (object): 4118 non-null values
- **change_type** (object): 4311 non-null values

### Data Statistics

**Total Records**: 4,311

**Migration Types**:
- function_signature_change: 4,118
- struct_layout_change: 193

**Change Types**:
- param_type_changed: 1,518
- return_type_changed: 1,448
- param_added: 789
- param_removed: 363
- fields_modified: 193

## Usage

```python
import pandas as pd

# Load the dataset
df = pd.read_parquet("hf://datasets/username/repo-name/migrations.parquet")

# Display first few rows
print(df.head())

# Filter by migration type
function_changes = df[df['type'] == 'function_signature_change']
print(f"Function signature changes: {len(function_changes)}")
```

## Data Collection

This dataset was generated using the `kernel-migrate` tool, which:
1. Fetches commits between specified kernel version tags
2. Parses C source and header files using tree-sitter
3. Detects API changes including function signature modifications and struct layout changes
4. Extracts metadata about each migration

## Citation

If you use this dataset in your research, please cite:

```bibtex
@dataset{kernel_migrations_v6.8_v6.9,
  title={Linux Kernel API Migrations: v6.8 to v6.9},
  author={Kernel Migrate Tool},
  year={2024}
}
```

## License

This dataset is derived from the Linux kernel source code, which is licensed under GPL-2.0.