Kernel Stack Trace Analyzer

A powerful tool for analyzing kernel stack traces using AI to help developers understand kernel issues and crashes.

Features

• Parse kernel stack traces from various sources
• Extract source code for each symbol in the call trace
• AI-powered analysis of stack traces
• Support for multiple stack traces and parallel analysis
• Context-aware analysis for different kernel error types:
  • Kernel Panic
  • Softlockup
  • KASAN issues (UAF, OOB)
  • Hung Tasks
• Root cause analysis and summarization

Installation

1. Clone the repository
2. Install dependencies:

pip install -r requirements.txt

3. Set up your OpenAI API key in a .env file:

OPENAI_API_KEY=your_api_key_here

or

export OPENAI_API_KEY=your_api_key_here

Usage

1. Prepare the linux kernel source code and build the cscope database.

cscope -Rbq

2. Run the analyzer:

python -m kernel_stack_analyzer --input panic.txt --context kernel_panic --kernel-src ~/linux-kernel-source/

or you can run with debug mode to get the prompt for other LLM models:

python -m kernel_stack_analyzer --input panic.txt --context kernel_panic --kernel-src ~/linux-kernel-source/ --debug

Project Structure

• kernel_stack_analyzer/
  • parser/ - Stack trace parsing modules
  • code_extractor/ - Source code extraction for symbols
  • ai_agents/ - AI analysis modules
  • context/ - Context-specific analyzers
  • utils/ - Utility functions
  • main.py - Main entry point