kdumpling

A Python library for creating Linux kdump crash dump files.

Overview

kdumpling allows you to synthesize valid ELF64 vmcore files from raw memory data and vmcoreinfo values. This is useful for:

• Testing crash dump analysis tools (like drgn, crash)
• Creating synthetic crash dumps for debugging
• Educational purposes for understanding Linux kernel crash dump formats

Installation

pip install kdumpling

Quick Start

from kdumpling import KdumpBuilder

# Create a builder for x86_64 architecture
builder = KdumpBuilder(arch='x86_64')

# Set the vmcoreinfo metadata
builder.set_vmcoreinfo("""OSRELEASE=5.14.0
PAGESIZE=4096
SYMBOL(swapper_pg_dir)=ffffffff82a00000
""")

# Add memory segments
builder.add_memory_segment(phys_addr=0x100000, data=b'\x00' * 4096)

# Add CPU register state (optional)
builder.add_cpu_context(
    cpu_id=0,
    registers={'RIP': 0xffffffff81000000, 'RSP': 0xffff888000000000},
    pid=1
)

# Check stats before writing
print(builder.stats)
# Dump Statistics:
#   Architecture: x86_64
#   Memory Segments: 1
#   CPU Contexts: 1
#   Total Memory: 4.0 KB (4096 bytes)
#   ...

# Write the vmcore file
builder.write("output.vmcore")

Features

• Multi-architecture support: x86_64, aarch64/arm64, s390x, ppc64/ppc64le, riscv64
• Fluent API: Chain method calls for concise code
• CPU context support: Include register state for debugging tools
• Memory from multiple sources: bytes, file paths, or file-like objects
• Statistics API: Inspect dump properties before writing
• Validated: Tested with pyelftools, drgn, and libkdumpfile

Supported Architectures

Validating with drgn

import drgn

prog = drgn.Program()
prog.set_core_dump("output.vmcore")
print(f"Platform: {prog.platform}")  # Platform(<Architecture.X86_64: 1>, ...)
print(f"Flags: {prog.flags}")        # ProgramFlags.IS_LINUX_KERNEL

API Reference

KdumpBuilder

builder = KdumpBuilder(arch='x86_64')

# Set vmcoreinfo metadata
builder.set_vmcoreinfo("OSRELEASE=5.14.0\n...")

# Add memory segments
builder.add_memory_segment(phys_addr=0x100000, data=b'...')
builder.add_memory_segment(phys_addr=0x200000, data="/path/to/file")

# Add CPU context
builder.add_cpu_context(cpu_id=0, registers={'RIP': 0x...}, pid=1)

# Get statistics
stats = builder.stats
print(stats.num_memory_segments)
print(stats.total_memory_size_human)

# Write to file
builder.write("output.vmcore")

DumpStats

stats = builder.stats

stats.architecture          # 'x86_64'
stats.num_memory_segments   # Number of PT_LOAD segments
stats.num_cpu_contexts      # Number of NT_PRSTATUS notes
stats.total_memory_size     # Total memory in bytes
stats.vmcoreinfo_size       # VMCOREINFO size in bytes
stats.estimated_file_size   # Estimated output file size
stats.memory_segments       # List of (phys_addr, size) tuples

# Human-readable sizes
stats.total_memory_size_human     # "4.0 MB"
stats.estimated_file_size_human   # "4.0 MB"

Development

# Clone the repository
git clone https://github.com/sdimitro/kdumpling.git
cd kdumpling

# Install development dependencies
pip install -e ".[dev]"

# Run tests
pytest

# Run linter
ruff check .

# Run type checker
mypy kdumpling

License

MIT License - see LICENSE for details.