Random freezes after last update (possibly amdgpu)

fernkaufmann

Hello Solus folks!

I'm running Solus on my AMD Ryzen 5 Pro 3400GE for more than 4 years now (Thinkcentre M75s Gen1). Never had any serious issues.

But since the last update, I encounter random freezes (Screen freezes, mouse cursor visible but not able to move). Every couple hours or so. While Browsing or in Idle.

Scrolling the logs wasn't very successful. Only dmesg shows some interesting messages, which may or may not be connected to the freezing.

When I run sudo dmesg -T -l err it shows:

[Mo, 30. Dez 2024, 15:00:21] amdgpu 0000:0d:00.0: amdgpu: Secure display: Generic Failure.
[Mo, 30. Dez 2024, 15:00:21] amdgpu 0000:0d:00.0: amdgpu: SECUREDISPLAY: query securedisplay TA failed. ret 0x0

And sudo dmesg -T -l warn shows:

[Mo, 30. Dez 2024, 15:00:16] Could not retrieve perf counters (-19)
[Mo, 30. Dez 2024, 15:00:21] amdgpu 0000:0d:00.0: amdgpu: psp gfx command LOAD_TA(0x1) failed and response status is (0x7)
[Mo, 30. Dez 2024, 15:00:21] amdgpu 0000:0d:00.0: amdgpu: psp gfx command INVOKE_CMD(0x3) failed and response status is (0x4)

What can I do to investigate further?

Even if the issue gets fixed with the next update, a sudden freeze could corrupt the update and ruin my system, I guess?

Romagnolo73

Welcome to the Ryzen world , I fight with those freez since the first day i bought a Ryzen laptop and installed linux distros. Same issue,
https://discuss.getsol.us/d/11228-how-to-load-amd-ucode-stored-on-boot
I can solve it using Arch but not Solus, on Arch I'm able to change amd-ucode.img and load a different one and I installed a C6-Disabler too available in AUR https://github.com/joakimkistowski/amd-disable-c6 but not available on Solus repo.
On Solus sadly I can't solve the issue so I use it only as "emergency" OS installed on other partition just in case my arch OS doesn't start.
And yes if freez happens updating OS can be a problematic situation, I agree.
On Solus you can only

try installing AMD microcode and pray there is a new version for your SoC.
wait for a new kernel hope it solve for you

tflovik

This is why Pop Os still is on kernel 6.9. They are waiting for a proper fix before they introduce a newer kernel.

Staudey

tflovik This is why Pop Os still is on kernel 6.9. They are waiting for a proper fix before they introduce a newer kernel.

Oh, do you have a link with info about that?

fernkaufmann

Thanks for the insights!

My Ryzen always did very well with the most current Kernel.
I somehow managed to install the LTS Kernel without a freeze. Testing it out now. Looking good so far.

tflovik

Staudey This is from the cosmic channel on Mattermost:
Generally, Pop!_OS does upgrade the kernel and Mesa to fairly recent versions, but newer kernels have some bugs (notably amdgpu), which might be why it hasn't yet been updated (there were pull requests for 6.10 and 6.11, but they weren't merged due to regressions).
As for Mesa, it's probably lower priority than COSMIC, and they might not yet have hardware that requires newer Mesa versions, so it's on 24.0.3.

SethStorm666

It must only affect Ryzen laptops as I'm on a Ryzen stationary and doesn't suffer such behavior.

storm@universe ~ $ inxi -Ca
CPU:
  Info: model: AMD Ryzen 9 5950X bits: 64 type: MT MCP arch: Zen 3+ gen: 3
    level: v3 note: check built: 2022 process: TSMC n6 (7nm) family: 0x19 (25)
    model-id: 0x21 (33) stepping: 2 microcode: 0xA201210
  Topology: cpus: 1x dies: 1 clusters: 1 cores: 16 threads: 32 tpc: 2
    smt: enabled cache: L1: 1024 KiB desc: d-16x32 KiB; i-16x32 KiB L2: 8 MiB
    desc: 16x512 KiB L3: 64 MiB desc: 2x32 MiB
  Speed (MHz): avg: 3776 min/max: 550/5084 boost: enabled scaling:
    driver: amd-pstate-epp governor: powersave cores: 1: 3776 2: 3776 3: 3776
    4: 3776 5: 3776 6: 3776 7: 3776 8: 3776 9: 3776 10: 3776 11: 3776 12: 3776
    13: 3776 14: 3776 15: 3776 16: 3776 17: 3776 18: 3776 19: 3776 20: 3776
    21: 3776 22: 3776 23: 3776 24: 3776 25: 3776 26: 3776 27: 3776 28: 3776
    29: 3776 30: 3776 31: 3776 32: 3776 bogomips: 217179
  Flags: avx avx2 ht lm nx pae sse sse2 sse3 sse4_1 sse4_2 sse4a ssse3

fernkaufmann

tflovik The release with kernel 6.10.13-304-1-x86_64 was the last one that worked for me.

fernkaufmann

No luck with newest LTS Kernel 6.6.66. Just had two freezes in a row 🙁

Any logs I could check? Anything else I could do besides waiting for bugfixes?

Romagnolo73

In BIOS setting you can see if your BIOS present a entry like GLOBAL-C STATE CONTROL or COOL&QUIET and disable it and try use your pc and see if this solve or not.
Generally disable C6 state solve issue on ZEN1 , ZEN+ and ZEN2 and you too is using a zen+ cpu as me
I don't have entry on BIOS on my Huawei laptop so the only solution was finding a distro with a C6 Disabler on the repo and is a ARCH based distro, but if you have the entry in your BIOS probably disabling C6 will solve and you can use Solos.
You can see if there is a BIOS update for your machine and if there is a update do it.
If the trick does nothing for you and still continuing freez then reactive C STATE on BIOS
In case it works this is simply a workaround, because one thing is only disabel C6 State (with a app or a setting after boot) and a different thing is disable C STATE at all as set it on the bios that cause your machine stay at maximum freq all the time. But if it solve and you have no more freez is better than nothing, and probably if needed we can see id Solus dev can insert a C6 disabler trick on the repo.

fernkaufmann

Romagnolo73 Thanks for the info! My BIOS is already up to date. I will look into the C State Bios settings.

fernkaufmann

It seems to be related to a bug which was introduced in Mesa 24.3.x. Interesting read: https://bbs.archlinux.org/viewtopic.php?id=301798

fernkaufmann

Someone found a way to reproduce this craziness:

I believe I've found a quick way to reproduce the issue. I use i3; I start two terminals in one workspace, and hold Super + Shift + Space on one of them, which rapidly alternates it between a floating window and a tiling window, producing a lot of graphical window resizing. This can produce a freeze on broken builds in a few seconds.

from https://bbs.archlinux.org/viewtopic.php?id=301798&p=3

TraceyC

Please open an issue (if there isn't one already) on our bug tracker for this. By opening an issue, it will allow our developers to look into it and keep track of progress. Thanks.