linux-kernel - Standby problem AMD Ryzen 9 and AMD GPU RX 6600

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite: Windows password security audit tool. GUI, reports in PDF.

[<prev] [next>] [day] [month] [year] [list]

Date:   Mon, 13 Mar 2023 20:55:33 +0100
From:   TuxCoder <coder@....ac>
To:     linux-kernel@...r.kernel.org
Subject: Standby problem AMD Ryzen 9 and AMD GPU RX 6600

Hello All,

This is my fist bugreport on the LKML so if I do something wrong please 
tell me.

Problem:

Since Kernel 6.0 the standby on my PC is not working properly.
The Error behavior is, that the PC tries to get into standby, goes for a 
brief moment to "power off" and starts right after that again.
after that the GPU is not working correctly, eg. the screen stays on but 
black, ctrl + alt + fX does nothing, (num lock is switching -> kernel 
alive).
SSH is still working.
Same root-fs system with a 5.19.0 does work.

System Description:

CPU: AMD Ryzen 9 5900X 12-Core
GPU: Radeon RX 6600
Mainboard: ASRock B550 Taichi Razer Edition
Distro: Debian Testing
Kernel-src: debian-testing / from kernel.org with makeold (and default 
for new values) from debian based, probably with small adjustments, can 
provide config if needed
GUI: Wayland + Sway from Debian testing
PCIE: Intel 10G Network Card: ASM1184e
....

As I understand enough from hardware and the GNU/Linux Kernel that power 
management is complicated, and I should not spam the LKML (FAQ attachments).
So I ask before I send all the logs and hardware information that I 
have. (please tell me what would be interesting)

As the core kernel does not crash, I can also debug / save the most part 
of the state over ssh, if needed from my laptop.
I have successful and crashes  logs dmesg of the older 5.19 and 6.0, 
6.1Kernel, see attachments
I also tested upstream Debian GNU/Linux Kernel from Testing is caching, 
to have an independent reference.

I attached an example suspend to ram and the following power up / crash 
of the AMD GPU driver as `amdgpu.fail`
also a success for `amdgpu.works`

I have also a bit knowledge about how to debug the GNU/Linux Kernel but 
not yet touched the GPU / power management driver part.

Could start a VM / reproduce this if needed, but currently not sure what 
I should debug / search for.

Thanks and best regards

TuxCoder

View attachment "amdgpu.fail" of type "text/plain" (241539 bytes)

View attachment "amdgpu.works" of type "text/plain" (131696 bytes)