Update 02/2024

The Problem has been solved by disabling TPM in UEFI Setup. TPM seems to be bugged with some CPUs. Before updating the BIOS, TPM was probably disabled by vendor, hence it didn’t appear initially.

If you experience freezes right before poweroff, try disabling TPM in UEFI/BIOS settings if it isn’t in use.

Symptoms

  • Complete Freeze after:
preparing to enter ACPI S5 state
Reboot: Power Down
  • seemlingly random (can successfully shutdown maybe 10 times in a row, then suddenly freezes again)
  • seems to happen more often as uptime grows
  • Case Fans still spin, LEDs and Lights stay on
  • Monitors stay on, still react to HDMI/DP Hotplugging (unplug/plug)
  • REISUB/REISUO doesn’t work
  • Disks are already powered off and disconnected
  • USB devices (eg. keyboard unresponsive)

Since When

  • After switching to Arch and flashing BIOS Firmware 7C88v18 to MSI B460M-A Pro Board
  • Arch Linux is ruled out by me since it’s very unlikely Userland plays a role, and already used Linux before without the problem
  • Gone from linux ~6.2.10 to 6.4.8, Kernel bug unlikely

Attempted:

  • Reflash 7C88v18 from a FAT32 formatted partition (USB)
  • Add reboot=pci or reboot=acpi acpi=force to kernel cmdline
  • run fwupd
  • Stop X and wait a bit for processes to clear up (???)
  • intel-ucode is installed:
~> pacman -Qi intel-ucode | head -2
Name                     : intel-ucode
Version                  : 20230613-1
  • Other threads on the Internet seem to have easier reproducibility (always happening), solutions were about either kernel cmdline or outdated kernel (If I didn’t see a thread with my exact problem, apologies)

Hardware Info

~> pacman -Qi nvidia | head -2
Name                     : nvidia
Version                  : 535.86.05-8

~> cat /sys/class/dmi/id/board_* 2>/dev/null
Default string
B460M-A PRO (MS-7C88)
Micro-Star International Co., Ltd.
1.0

~> LC_ALL=C lscpu | grep -i 'model name'
Model name:                      Intel(R) Core(TM) i7-10700 CPU @ 2.90GHz

~> sudo journalctl -k --grep=microcode
[...] kernel: microcode: updated early: 0xf0 -> 0xf6, date = 2022-12-26
[...] kernel: SRBDS: Mitigation: Microcode
[...] kernel: microcode: Microcode Update Driver: v2.2.

Misc

  • Did I just miss something obvious?
  • I don’t want to go back to an outdated BIOS firmware
  • lscpu, nvidia-smi and other info added if needed
  • I have another device (ASUS Board, similar CPU) with similar arch linux setup (nouveau instead of propietary nvidia there), no problems on that device
  • BaalInvokerA
    2·
    1 年前

    I know you said that is unlikely that is a kernel bug, however have you tried kernel-lts?

    I had many issues with newer kernels and lts saved me