[震惊] 用 Firefox 播放一个抖音视频，我的 Linux 系统立马死机

由于我的系统是魔改的 Ubuntu 16.04 + Kernel 6.19 + Mesa 25.1.4 + libva 2.11 + Firefox 152(参考： https://v2ex.com/t/1179778 ), 最近突然发现机器经常卡顿，最终定位发现播放某一个抖音视频（《归墟》 第一季 1-9 合集）时，立马出发 AMDGPU VCNU 固件报错，严重时显卡失去响应，需要强制断电重启。

嗯，这也算一种 Linux “攻击”吧😁，通过某些特殊的视频编码，层层穿透，直达内核。

打算切换到 nix 的 mesa 26.1.5 + Firefox 152 了。

OS: Ubuntu 16.04.7 LTS (Xenial Xerus) x86_64
Host: 21LF (ThinkBook 14 G6+ AHP)
Kernel: Linux 6.19.14-zabbly+
Uptime: 11 mins
Packages: 1112 (dpkg), 807 (nix-user)
Shell: bash 4.3.48
Display (LEN8AAF): 3072x1920 @ 1.75x in 14", 120 Hz *
Display (DELL P2722H): 1920x1080 @ 1.75x in 27", 60 Hz [External]
DE: Unity 7.4.5
WM: Compiz (X11)
Theme: MacBuntu-OS-Unity-Classic [GTK2/3/4]
Icons: Papirus-Light [GTK2/3/4]
Font: Ubuntu (11pt) [GTK2/3/4]
Cursor: DMZ-White (21px)
Terminal: GNOME Terminal 3.18.3
Terminal Font: Monospace (13pt)
CPU: AMD Ryzen 7 8845H (16) @ 5.14 GHz
GPU: AMD Radeon 780M Graphics [Integrated]

amdgpu 0000:64:00.0: amdgpu: MMVM_L2_PROTECTION_FAULT_STATUS:0x00F03A30
amdgpu 0000:64:00.0: amdgpu: 	 Faulty UTCL2 client ID: VCN (0x1d)
amdgpu 0000:64:00.0: amdgpu: 	 MORE_FAULTS: 0x0
amdgpu 0000:64:00.0: amdgpu: 	 WALKER_ERROR: 0x0
amdgpu 0000:64:00.0: amdgpu: 	 PERMISSION_FAULTS: 0x3
amdgpu 0000:64:00.0: amdgpu: 	 MAPPING_ERROR: 0x0
amdgpu 0000:64:00.0: amdgpu: 	 RW: 0x0
amdgpu 0000:64:00.0: amdgpu: [mmhub] page fault (src_id:0 ring:24 vmid:15 pasid:25)
amdgpu 0000:64:00.0: amdgpu:  Process RDD Process pid 2343 thread browser 4 :cs0 pid 2864
amdgpu 0000:64:00.0: amdgpu:   in page starting at address 0x0000800102ab9000 from client 18
amdgpu 0000:64:00.0: amdgpu: MMVM_L2_PROTECTION_FAULT_STATUS:0x00F03A30
amdgpu 0000:64:00.0: amdgpu: 	 Faulty UTCL2 client ID: VCN (0x1d)
amdgpu 0000:64:00.0: amdgpu: 	 MORE_FAULTS: 0x0
amdgpu 0000:64:00.0: amdgpu: 	 WALKER_ERROR: 0x0
amdgpu 0000:64:00.0: amdgpu: 	 PERMISSION_FAULTS: 0x3
amdgpu 0000:64:00.0: amdgpu: 	 MAPPING_ERROR: 0x0
amdgpu 0000:64:00.0: amdgpu: 	 RW: 0x0
amdgpu 0000:64:00.0: amdgpu: [mmhub] page fault (src_id:0 ring:24 vmid:15 pasid:25)
amdgpu 0000:64:00.0: amdgpu:  Process RDD Process pid 2343 thread browser 4 :cs0 pid 2864
amdgpu 0000:64:00.0: amdgpu:   in page starting at address 0x0000800102ab9000 from client 18
amdgpu 0000:64:00.0: amdgpu: MMVM_L2_PROTECTION_FAULT_STATUS:0x00F03A30
amdgpu 0000:64:00.0: amdgpu: 	 Faulty UTCL2 client ID: VCN (0x1d)
amdgpu 0000:64:00.0: amdgpu: 	 MORE_FAULTS: 0x0
amdgpu 0000:64:00.0: amdgpu: 	 WALKER_ERROR: 0x0
amdgpu 0000:64:00.0: amdgpu: 	 PERMISSION_FAULTS: 0x3
amdgpu 0000:64:00.0: amdgpu: 	 MAPPING_ERROR: 0x0
amdgpu 0000:64:00.0: amdgpu: 	 RW: 0x0
amdgpu 0000:64:00.0: amdgpu: Dumping IP State
amdgpu 0000:64:00.0: amdgpu: Dumping IP State Completed
amdgpu 0000:64:00.0: amdgpu: [drm] AMDGPU device coredump file has been created
amdgpu 0000:64:00.0: amdgpu: [drm] Check your /sys/class/drm/card0/device/devcoredump/data
amdgpu 0000:64:00.0: amdgpu: ring vcn_unified_0 timeout, signaled seq=28, emitted seq=30
amdgpu 0000:64:00.0: amdgpu:  Process RDD Process pid 2343 thread browser 4 :cs0 pid 2864
amdgpu 0000:64:00.0: amdgpu: Starting vcn_unified_0 ring reset
amdgpu 0000:64:00.0: amdgpu: Register(0) [regUVD_POWER_STATUS] failed to reach value 0x00000001 != 0x00000002n
amdgpu 0000:64:00.0: amdgpu: Register(0) [regUVD_RB_RPTR] failed to reach value 0x00000200 != 0x00000180n
amdgpu 0000:64:00.0: amdgpu: Register(0) [regUVD_POWER_STATUS] failed to reach value 0x00000001 != 0x00000002n
amdgpu 0000:64:00.0: amdgpu: Ring vcn_unified_0 reset succeeded
amdgpu 0000:64:00.0: [drm] device wedged, but recovered through reset