====== 🛡️ System Crash & Recovery Reports ====== This page documents all system freezes, hard lockups, and kernel panics, including the analysis of logs captured post-recovery. ===== 📅 2026-03-21: REISUB (Hard Lockup) ===== ^ Field ^ Details ^ | **Type** | Hard Lockup (Host) | | **Trigger** | Magic SysRq (REISUB) | | **Approx Time** | 14:12:00 | | **Last Journal Entry** | ''kernel: vfio-pci 0000:00:02.1: reset done'' | ==== 🔍 Root Cause Analysis ==== 1. **iGPU VF Reset**: The crash occurred immediately after the VM attempted to enable and reset the iGPU VF (00:02.1). The host logs show ''VF1 FLR'' (Function Level Reset) followed by ''reset done'', but the journal stopped abruptly after this. 2. **Process Noise**: The ''python3'' process (Gemini Voice) was in a high-frequency loop logging ''"Error: API Key not found"'' every 5 seconds. This contributed to high I/O and CPU context switching overhead during the sensitive VF reset period. 3. **Service Failure**: ''max-performance.service'' had a fatal configuration error, potentially leaving the CPU/GPU in an inconsistent power state before the VF was engaged. ==== 🛠️ Remediation Applied ==== * **Disabled Gemini Voice**: Persistent removal of the service to eliminate the high-frequency error loop. * **Fixed Performance Service**: Refactored into a standalone script to ensure stable power states on boot. * **Btrfs Check**: Verified Btrfs integrity after the unclean shutdown (renamed/replaced corrupted journal). ==== 📈 Outcome ==== System stabilized after post-reboot optimizations and removal of the failing voice assistant process. No further hangs detected during VM XML updates.