Complete freeze linux of AMD Ryzen 7 Radeon 680M

zayn7lie · 2023-08-17 00:46:05

[Redirection/Solution]

A more detailed discussion could be found here: https://gitlab.freedesktop.org/drm/amd/-/issues/2068

Updated: the problem exists also in Windows, it might relate to hardware problem.

[Bug Description]

The screen and system will totally freeze randomly without sync to dmesg or journalctl. This situation both happens within DE/WM and console (original tty), and both current kernel and lts kernel.

[Log]

N/A (No log while system freeze)

[System Info]

$ neofetch
                   -`                    zayn7lie@zayn7lie-acl 
                  .o+`                   --------------------- 
                 `ooo/                   OS: Arch Linux x86_64 
                `+oooo:                  Host: ROG Flow X13 GV301RA_GV301RA 1.0 
               `+oooooo:                 Kernel: 6.6.16-1-lts 
               -+oooooo+:                Uptime: 14 mins 
             `/:-:++oooo+:               Packages: 559 (pacman) 
            `/++++/+++++++:              Shell: bash 5.2.26 
           `/++++++++++++++:             Resolution: 1920x1200 
          `/+++ooooooooooooo/`           WM: Hyprland 
         ./ooosssso++osssssso+`          Terminal: kitty 
        .oossssso-````/ossssss+`         CPU: AMD Ryzen 7 6800HS with Radeon Graphics (16) @ 4.785GHz 
       -osssssso.      :ssssssso.        GPU: AMD ATI Radeon 680M 
      :osssssss/        osssso+++.       Memory: 1714MiB / 15230MiB 
     /ossssssss/        +ssssooo/-       GPU Driver: amdgpu 
   `/ossssso+/:-        -:/+osssso+-     CPU Usage: 2% 
  `+sso+:-`                 `.-/+oso:    Disk (/): 60G / 251G (25%) 
 `++:.                           `-/+/   Locale: C.UTF-8 
 .`                                 `/

[misc]

I have changed my laptop and closed the thread. Please go to [Redirection] for further research.

Last edited by zayn7lie (2024-03-07 18:23:52)

seth · 2023-08-17 08:07:18

The kernel halts, the journal is "useless" because you're somewhat forced to reboot w/ the power button and it doesn't get synced to the disk.
Try https://wiki.archlinux.org/title/Keyboa … el_(SysRq) and monitor "dmesg -W" in a visible terminal, leading up to the halt.

Since it's a ryzen system, see https://wiki.archlinux.org/title/Ryzen#Troubleshooting
Notably the curve optimiser to increase the CPU voltage and "processor.max_cstate=1"

zayn7lie · 2023-08-17 09:33:27

seth wrote:

The kernel halts, the journal is "useless" because you're somewhat forced to reboot w/ the power button and it doesn't get synced to the disk.
Try https://wiki.archlinux.org/title/Keyboa … el_(SysRq) and monitor "dmesg -W" in a visible terminal, leading up to the halt.
Since it's a ryzen system, see https://wiki.archlinux.org/title/Ryzen#Troubleshooting
Notably the curve optimiser to increase the CPU voltage and "processor.max_cstate=1"

Unfortunately, the ROG laptop has not "Sysrq" key.

Thanks, I have limited my cstate to 1 and I will test if this really work.

$ journalctl -k --grep=processor
Aug 17 05:26:31 zayn7lie-acl kernel: Command line: BOOT_IMAGE=/vmlinuz-linux root=UUID=1794e657-d6a0-446c-9067-294f7198fbb0 rw loglevel=3 quiet processor.max_cstate=1
Aug 17 05:26:31 zayn7lie-acl kernel: tsc: Detected 3194.140 MHz processor
Aug 17 05:26:31 zayn7lie-acl kernel: Kernel command line: BOOT_IMAGE=/vmlinuz-linux root=UUID=1794e657-d6a0-446c-9067-294f7198fbb0 rw loglevel=3 quiet processor.max_cstate=1
Aug 17 05:26:31 zayn7lie-acl kernel: smpboot: Total of 16 processors activated (102253.33 BogoMIPS)
Aug 17 05:26:31 zayn7lie-acl kernel: ACPI: Added _OSI(Processor Device)
Aug 17 05:26:31 zayn7lie-acl kernel: ACPI: Added _OSI(Processor Aggregator Device)
Aug 17 05:26:32 zayn7lie-acl kernel: ACPI: processor limited to max C-state 1

seth · 2023-08-17 12:32:56

It's likely your PrtScn (print screen) key.

zayn7lie · 2023-08-18 10:11:34

`processor.max_cstate=1` only reduce the frequency but not solve it totally

Last edited by zayn7lie (2023-08-22 04:18:19)

zayn7lie · 2023-08-23 09:14:42

Updated. After removing `power-profiles-daemon` or change from performance to balance and quiet mode, the random freezing would change into random reboot.

BTW, what is the difference between soft freeze and hard freeze?

Last edited by zayn7lie (2023-08-23 09:35:40)

seth · 2023-08-23 11:40:16

seth wrote:

Notably the curve optimiser to increase the CPU voltage and "processor.max_cstate=1"

I assume you mean the menitioned soft lockups?
https://docs.kernel.org/admin-guide/loc … hdogs.html

zayn7lie · 2023-09-03 13:54:01

FINALLY, I think I found the correct kernel parameter `amdgpu.dpm=1` (dynamic power management setting (0 = disable, 1 = enable) The default is -1 (auto).)

Ref: https://gitlab.freedesktop.org/drm/amd/-/issues/2068 There seems to me lots of discussion about dpm.

Audio loop is not special while system completely freeze so never mind.

seth · 2023-09-03 14:04:25

Have you tried it?
By recent accounts it tended to completely prevent the GPU from initiating (and I stopped suggesting it, so if that's no longer the case…)

zayn7lie · 2023-09-03 15:51:06

seth wrote:

Have you tried it?
By recent accounts it tended to completely prevent the GPU from initiating (and I stopped suggesting it, so if that's no longer the case…)

I have tried it for days. Cause it is very hard to reproduce directly, I would report it if there is other issue.

seth · 2023-09-03 17:46:59

I just meant whether you're running with this parameter at all, because then now it's back on the top spot of what to try if your AMD GPU drops out
Many thanks for the heads up!

zayn7lie · 2023-09-10 22:55:52

Unfortunately, the parameter just greatly reduce the frequency, the freezing would still happen sometimes. (But the frequency of freezing is bearable now)

Here are the parameters I have already tested:
amdgpu.runpm=0
amd.ppfeaturemask=0xffffbffb OR 0xfffffffe OR 0
amd_pstate=passive OR active
amdgpu.dpm=1
amdgpu.vm_update_mode=3
processor.max_cstate=1/5
iommu=pt

Current:
amdgpu.vm_update_mode=3 amdgpu.dcdebugmask=0x4

[Solved] deleted. Looking forward to more solution~

Ref: https://gitlab.freedesktop.org/drm/amd/-/issues/2068

Last edited by zayn7lie (2024-02-15 04:58:46)

seth · 2023-09-11 06:14:07

Ok, I kinda overlooked something … "amdgpu.dpm=1" isn't what I jumped into my mind and is completely inert, dpm is active by default anyway.

zayn7lie · 2024-02-16 15:54:10

Here are some updates:
When I limit processor.max_cstate=0 or 1 as a kernel parameter, the freezing will change to reboot. But if I limit it to 2 or 3, the freezing is still exist.
But neither of them were recorded in dmesg or journalctl.
I have disable tlp for a while, but it seems to be more stable when I used tlp with AC.

Last edited by zayn7lie (2024-02-16 15:58:25)

dav1ta · 2024-02-20 15:11:48

I had similar problem. tried everything above but nothing worked.
in the end problem was in cpu frequency governor.
i changed it from cpufreq to amd-pstate and problem were gone.
check out this wiki. https://wiki.archlinux.org/title/CPU_frequency_scaling

Arch Linux

#1 2023-08-17 00:46:05

Complete freeze linux of AMD Ryzen 7 Radeon 680M

#2 2023-08-17 08:07:18

Re: Complete freeze linux of AMD Ryzen 7 Radeon 680M

#3 2023-08-17 09:33:27

Re: Complete freeze linux of AMD Ryzen 7 Radeon 680M

#4 2023-08-17 12:32:56

Re: Complete freeze linux of AMD Ryzen 7 Radeon 680M

#5 2023-08-18 10:11:34

Re: Complete freeze linux of AMD Ryzen 7 Radeon 680M

#6 2023-08-23 09:14:42

Re: Complete freeze linux of AMD Ryzen 7 Radeon 680M

#7 2023-08-23 11:40:16

Re: Complete freeze linux of AMD Ryzen 7 Radeon 680M

#8 2023-09-03 13:54:01

Re: Complete freeze linux of AMD Ryzen 7 Radeon 680M

#9 2023-09-03 14:04:25

Re: Complete freeze linux of AMD Ryzen 7 Radeon 680M

#10 2023-09-03 15:51:06

Re: Complete freeze linux of AMD Ryzen 7 Radeon 680M

#11 2023-09-03 17:46:59

Re: Complete freeze linux of AMD Ryzen 7 Radeon 680M

#12 2023-09-10 22:55:52

Re: Complete freeze linux of AMD Ryzen 7 Radeon 680M

#13 2023-09-11 06:14:07

Re: Complete freeze linux of AMD Ryzen 7 Radeon 680M

#14 2024-02-16 15:54:10

Re: Complete freeze linux of AMD Ryzen 7 Radeon 680M

#15 2024-02-20 15:11:48

Re: Complete freeze linux of AMD Ryzen 7 Radeon 680M

Board footer