Modernizing the kdump dump tools
summary
The kdump toolchain (makedumpfile + crash) accumulates decades of unstable KABI parsing, causing real bugs and security risk — one sparse-mem change from 2017 led to dump corruption on s390 and required six engineers and 10 months to fix. Proposed solution is to stop supporting every old kernel: reduce KABI versions, extend BTF or ship debugger metadata, and kdump tools moved into kernel tree so tools follow kernel stable/backport workflows. Migration requires rewriting crash (GDB dependency → Python API), multi‑year support overlap, and distro packaging changes.
AI-written summary · about
chapters (18)
- 0:00Introduction and talk overview
- 1:08What is kdump and its tools
- 3:16Problems with old tools design
- 7:39Security and complexity issues
- 10:31Options to modernize kdump tools
- 14:49Moving tools into kernel tree
- 17:02Q&A on security and isolation
- 19:32Q&A on kernel version support
- 22:48BTF and crash tool redesign
- 26:22Distro and vendor challenges
- 28:30LTS kernel backward compatibility
- 30:16Alternative debuggers like dragon
- 33:33Crash Python and automation
- 36:23Upstream and downstream problems
- 38:31Packaging and distribution
- 41:23Future of kernel self-description
- 42:48Modularizing crash tool
- 44:57Closing remarks and Q&A summary
related talks
49mKernel dump analysis made quick and easy!
20mDebugging RISC-V Kernel Crashes: KDUMP, vmcore, and Crash Utility Enhancements
19mKdump memory usage
37mPracticing Linux Crash/Panic Issue on Production and Cloud Server: Using Kdump + Crash
13mRamdump for Trusted VMs: Efficient Crash Debugging Without Reboot Device
52mLinux Kernel Debugging: Going Beyond Printk Messages
37mHardware Error Handling Improvement for Reliable KVM Hypervisor
43mEfficient and Practical Capturing of Crash Data on Embedded Systems