Proactive and crash-time data collection on Steam Deck
summary
Steam Deck collects both proactive metrics and crash-time artifacts using pstore default (≈60 MB RAM) for fast persistent logs and kdump for full vmcore when deeper state is needed; kdump gives full memory snapshots but adds reserved-memory, size and privacy tradeoffs. Tracing captures split-lock and MDS-like hazards via trace_pipe, GPU traces and mini/core dumps capture graphics failures, and panic notifiers run extra collectors (current work splits and hardens them). An SLS agent aggregates artifacts and submits to Valve Sentry on opt-in, with DRM/UEFI panic paths still work-in-progress.
AI-written summary · about
chapters (8)
- 0:00Introduction to data collection
- 1:27Steam Deck overview and goals
- 4:59Crash data collection methods
- 8:54Pstore vs Kdump trade-offs
- 13:26GPU reset and failure handling
- 14:31Metrics and performance data
- 17:44Data aggregation with SLS
- 20:34Q&A and closing remarks
related talks
27mWhen kdump is way too much
34mPostmortem Debugging with Coresight
43mEfficient and Practical Capturing of Crash Data on Embedded Systems
28mMinidump to debug end user device crashes
52mLinux Kernel Debugging: Going Beyond Printk Messages
37mPracticing Linux Crash/Panic Issue on Production and Cloud Server: Using Kdump + Crash
48mGraphics Performance Analysis with FrameRetrace: A Responsive UI for ApiTrace
49mKernel dump analysis made quick and easy!