[Solved] kacpi_notify eating all cpu's 1st core

Linux Kernel, Network, and Services configuration.
Post Reply
Message
Author
sequencer
Posts: 29
Joined: 2024-03-29 17:16
Has thanked: 4 times

[Solved] kacpi_notify eating all cpu's 1st core

#1 Post by sequencer »

kworker/0:1+kacpi_notify eating all cpu 1st core
17 root 20 R 100.0 58:45.55 `- [kworker/0:1+kacpi_notify]

I'm running 6.1.0-25-amd64. (Debian 12)

Code: Select all

dmesg | grep -i error
[    0.252655] ACPI BIOS Error (bug): Could not resolve symbol [\_SB.PCI0.GPP6.WLAN], AE_NOT_FOUND (20220331/dswload2-162)
[    0.252659] ACPI Error: AE_NOT_FOUND, During name lookup/catalog (20220331/psobject-220)
[    0.253322] ACPI BIOS Error (bug): Could not resolve symbol [\_SB.PCI0.GPP2.WWAN], AE_NOT_FOUND (20220331/dswload2-162)
[    0.253324] ACPI Error: AE_NOT_FOUND, During name lookup/catalog (20220331/psobject-220)
[    0.253332] ACPI BIOS Error (bug): Could not resolve symbol [\_SB.PCI0.GPP5.RTL8], AE_NOT_FOUND (20220331/dswload2-162)
[    0.253334] ACPI Error: AE_NOT_FOUND, During name lookup/catalog (20220331/psobject-220)
[    0.253347] ACPI BIOS Error (bug): Failure creating named object [\_SB.PCI0.GPP6._PRW], AE_ALREADY_EXISTS (20220331/dswload2-326)
[    0.253349] ACPI Error: AE_ALREADY_EXISTS, During name lookup/catalog (20220331/psobject-220)
[    0.253354] ACPI BIOS Error (bug): Could not resolve symbol [\_SB.PCI0.GPP6.DEV0], AE_NOT_FOUND (20220331/dswload2-162)
[    0.253356] ACPI Error: AE_NOT_FOUND, During name lookup/catalog (20220331/psobject-220)
[    2.025253] amdgpu 0000:05:00.0: Direct firmware load for amdgpu/gc_11_0_1_mes_2.bin failed with error -2

Code: Select all

journalctl -k | grep acpi
Sep 18 07:54:47 omen kernel: ACPI: LAPIC_NMI (acpi_id[0xff] high edge lint[0x1])
Sep 18 07:54:47 omen kernel: acpiphp: ACPI Hot Plug PCI Controller Driver version: 0.5
Sep 18 07:54:47 omen kernel: acpi PNP0A08:00: _OSC: OS supports [ExtendedConfig ASPM ClockPM Segments MSI HPX-Type3]
Sep 18 07:54:47 omen kernel: acpi PNP0A08:00: _OSC: OS now controls [PCIeHotplug SHPCHotplug PME AER PCIeCapability LTR]
Sep 18 07:54:47 omen kernel: acpi PNP0A08:00: FADT indicates ASPM is unsupported, using BIOS configuration
Sep 18 07:54:47 omen kernel: clocksource: acpi_pm: mask: 0xffffff max_cycles: 0xffffff, max_idle_ns: 2085701024 ns
Sep 18 07:54:48 omen kernel: acpi-tad ACPI000E:00: Missing _PRW
Sep 18 07:54:53 omen kernel: ucsi_acpi USBC000:00: PPM init failed (-5)
any help appreciated

P.S.: can't find information about top displaying cpus values. but I guess 0/100 100 means user processes utilize 0, system processes (root user) 100% utilization, and last is total. (0/100 also have different colors on bar)
Last edited by sequencer on 2024-09-24 20:05, edited 1 time in total.

User avatar
stevepusser
Posts: 13023
Joined: 2009-10-06 05:53
Has thanked: 45 times
Been thanked: 91 times

Re: kacpi_notify eating all cpu's 1st core

#2 Post by stevepusser »

Is your mystery machine using the latest firmware/UEFI from its maker?

You could also try a newer kernel to see if it has added quirks for your problem.
MX Linux packager and developer

sequencer
Posts: 29
Joined: 2024-03-29 17:16
Has thanked: 4 times

Re: kacpi_notify eating all cpu's 1st core

#3 Post by sequencer »

Installed kernel 6.10.9 from sid. And now it's creating 100x kacpi_notify on random core. Where each process takes 2% and in combination 100% one core always.
I've updated UEFI. Didn't help.

P.S.: I guess now it's not root user utilization, but kernel's.

Aki
Global Moderator
Global Moderator
Posts: 3949
Joined: 2014-07-20 18:12
Location: Europe
Has thanked: 109 times
Been thanked: 518 times

Re: kacpi_notify eating all cpu's 1st core

#4 Post by Aki »

Hello,

Take a look at here to check for acpi irq involved (if any): Does the situation improve disconnecting all connected external peripherals (e.g.: usb hubs, usb C connected devices, etc) ?
⢀⣴⠾⠻⢶⣦⠀
⣾⠁⢠⠒⠀⣿⡁ Debian - The universal operating system
⢿⡄⠘⠷⠚⠋⠀ https://www.debian.org
⠈⠳⣄⠀

sequencer
Posts: 29
Joined: 2024-03-29 17:16
Has thanked: 4 times

Re: kacpi_notify eating all cpu's 1st core

#5 Post by sequencer »

Aki wrote: 2024-09-19 20:08
Hello. No, removing everything from my laptop's ports aside power connector didn't help. Before I had only my USB-A mouse connected.
I don't quite actually understand the discussion in the kernel bugzilla you provided. Can you prove my point please, these 0/100 100 means core is occupied by some kernel processes? I don't actually feel my laptop stressed by one random core running wild but I have concerns about my cpu longevity.

User avatar
sunrat
Site admin
Site admin
Posts: 7365
Joined: 2006-08-29 09:12
Location: Melbourne, Australia
Has thanked: 133 times
Been thanked: 649 times

Re: kacpi_notify eating all cpu's 1st core

#6 Post by sunrat »

sequencer wrote: 2024-09-19 19:14 Installed kernel 6.10.9 from sid.
You said you are on Debian 12. You should not install kernels from Sid!
If you need a newer kernel, install it from bookworm-backports.
“ computer users can be divided into 2 categories:
Those who have lost data
...and those who have not lost data YET ”
Remember to BACKUP!

sequencer
Posts: 29
Joined: 2024-03-29 17:16
Has thanked: 4 times

Re: kacpi_notify eating all cpu's 1st core

#7 Post by sequencer »

sunrat wrote: 2024-09-19 23:37
I tried both. mentioned latest

Aki
Global Moderator
Global Moderator
Posts: 3949
Joined: 2014-07-20 18:12
Location: Europe
Has thanked: 109 times
Been thanked: 518 times

Re: kacpi_notify eating all cpu's 1st core

#8 Post by Aki »

sequencer wrote: 2024-09-19 21:37 these 0/100 100 means core is occupied by some kernel processes?
Yes, the kacpi_notify is what is known as a "kernel workqueue", a sort of "kernel task" that is woken up by your computer's ACPI firmware (BIOS) every time the ACPI firmware notifies an event to the operating system (Linux kernel).

This is another example about a similar discussion (with similar ACPI error messages in a HP notebook and probably a similar ACPI kernel module involved named ucsi_acpi):
⢀⣴⠾⠻⢶⣦⠀
⣾⠁⢠⠒⠀⣿⡁ Debian - The universal operating system
⢿⡄⠘⠷⠚⠋⠀ https://www.debian.org
⠈⠳⣄⠀

sequencer
Posts: 29
Joined: 2024-03-29 17:16
Has thanked: 4 times

Re: kacpi_notify eating all cpu's 1st core

#9 Post by sequencer »

And neither was the problem solved there. 🥲
I should probably participate in conversation there. Thank you Aki!

sequencer
Posts: 29
Joined: 2024-03-29 17:16
Has thanked: 4 times

Re: kacpi_notify eating all cpu's 1st core

#10 Post by sequencer »

I have added module to black list as in https://bugs.launchpad.net/ubuntu/+sour ... comments/3

Code: Select all

GRUB_CMDLINE_LINUX_DEFAULT="quiet splash module_blacklist=ucsi_acpi"
And it fixed issue. But can we call this a "fix"? Or is it just temporary workaround. I didn't find any differences yet but I don't how does this affect my system.

Aki
Global Moderator
Global Moderator
Posts: 3949
Joined: 2014-07-20 18:12
Location: Europe
Has thanked: 109 times
Been thanked: 518 times

Re: kacpi_notify eating all cpu's 1st core

#11 Post by Aki »

Hello,

I'm glad I helped you to sort it out.
sequencer wrote: 2024-09-20 10:18 it fixed issue. But can we call this a "fix"? Or is it just temporary workaround.
Probably it does not matter the definition. You are running 6.1.0-25-amd64 and the ucsi_acpi kernel module [1]. You can test a newer kernel versions. If the issue is still there, you can open a bug report on the Debian Bug Tracking System (https://bugs.debian.org) and/or notify it to the Linux kernel maintainer of the kernel module.

Some changes has been made to the ucsi_acpi.c in latest kernel versions; these are the commits:

Code: Select all

$ git log --format="%h %as %s" v6.1.106..v6.10.9  -- drivers/usb/typec/ucsi/ucsi_acpi.c 
9e3caa9dd51b 2024-06-12 usb: typec: ucsi_acpi: Add LG Gram quirk
4ed37ef8678d 2024-03-27 usb: typec: ucsi_acpi: Remove Dell quirk
f3031f9b39c1 2024-03-27 usb: typec: ucsi: Stop abuse of bit definitions from ucsi.h
6aaceb7d9cd0 2024-03-20 usb: typec: ucsi_acpi: Refactor and fix DELL quirk
f3be347ea42d 2024-01-21 usb: ucsi_acpi: Quirk to ack a connector change ack cmd
2840143e393a 2024-01-21 usb: ucsi_acpi: Fix command completion handling
fc4ecc0cd561 2023-05-18 usb: typec: ucsi: acpi: Convert to platform remove callback returning void
326e1c208f3f 2023-04-05 usb: typec: ucsi: acpi: add quirk for ASUS Zenbook UM325
02d210f43424 2023-03-08 usb: ucsi_acpi: Increase the command completion timeout
1fab45ab6e82 2022-12-12 Merge tag 'rcu.2022.12.02a' of git://git.kernel.org/pub/scm/linux/kernel/git/paulmck/linux-rcu
Therefore, you may consider testing a latest kernel version.
sequencer wrote: 2024-09-20 10:18 I didn't find any differences yet but I don't how does this affect my system.
The ucsi_acpi kernel module is not playing well with your USB Type-C™ Connector System Software Interface [UCSI] [2] ACPI firmware.

Perhaps some ACPI power management events for USB Type-C™ connected devices are not notified to the kernel when ucsi_acpi.ko kernel module is disabled.

Probably not a big deal.

--
[1] sources / linux / 6.1.106-3 / drivers / usb / typec / ucsi / ucsi_acpi.c
[2] USB Type-C™ Connector System Software Interface [UCSI]
⢀⣴⠾⠻⢶⣦⠀
⣾⠁⢠⠒⠀⣿⡁ Debian - The universal operating system
⢿⡄⠘⠷⠚⠋⠀ https://www.debian.org
⠈⠳⣄⠀

Post Reply