Postmortem Debugging with Coresight
summary
Shows how to get actual program-counter traces for postmortem analysis: uses Coresight CPU debug module to read live CPU PC registers for lockup cases and uses Coresight panic kdump plus preserved ETM/CTI in DDR to reconstruct the last execution flow. Demonstrates on a 96Boards Hikey, extracts trace blobs, converts them to perf-compatible files and decodes with perf + OpenCST to produce a symbolized call stream. Includes a working demo of pstore/pmsg fallback and a 25-minute end-to-end kdump + trace extraction.
AI-written summary · about
chapters (9)
- 0:00Introduction to postmortem debugging
- 2:14Debugging system hang issues
- 8:55Security state debugging
- 12:38CPU state detection demo
- 15:10Hardware-assisted tracing overview
- 19:47Coresight tracing setup demo
- 24:55Decoding trace data with perf
- 29:26Extending the debugging workflow
- 30:11Q&A session
related talks
52mLinux Kernel Debugging: Going Beyond Printk Messages
37mPracticing Linux Crash/Panic Issue on Production and Cloud Server: Using Kdump + Crash
55mLinux Kernel Debug And Profiling Tools
43mEfficient and Practical Capturing of Crash Data on Embedded Systems
49mKernel dump analysis made quick and easy!
31mFantastic tracepoints and where to find them
34mBridging the Gap Between Hardware and Software Tracing
20mDebugging RISC-V Kernel Crashes: KDUMP, vmcore, and Crash Utility Enhancements