Debugging RISC-V Kernel Crashes: KDUMP, vmcore, and Crash Utility Enhancements
summary
KDUMP on a real VisionFive2 board needs specific fixes: applying two patches enables actual vmcore extraction (author collects >10 vmcores), then loading those dumps reveals wrong page-table and CSR displays in the crash utility; a patch set corrects the V2→V2P output and gets merged upstream. The work references roughly 30 architecture patches upstreamed overall with about 15–16 changes touching riscv64.c, and the extraction-plus-analysis flow now produces usable crash snapshots for offline debugging.
AI-written summary · about
chapters (10)
- 0:00Introduction and background
- 1:49Debugging challenges in RISC-V
- 3:17VMcore RAM dump importance
- 5:07KDump and Crash utility overview
- 8:40Case study and crash analysis
- 9:25KDump challenges on VisionFive2
- 10:30Patches for VMcore extraction
- 11:35Crash utility output issues
- 13:25Future work and upstreaming patches
- 15:15Q&A session
related talks
49mKernel dump analysis made quick and easy!
45mModernizing the kdump dump tools
36mRISC-V Bug Hunting
52mLinux Kernel Debugging: Going Beyond Printk Messages
34mPostmortem Debugging with Coresight
37mHardware Error Handling Improvement for Reliable KVM Hypervisor
43mEfficient and Practical Capturing of Crash Data on Embedded Systems
27mTuning in-kernel routines on RISC-V