amdgpu: graphical output borken since >=linux-6.7.6
Before submitting your bug report:
I've checked if my bug was already reported but couldn't find one. However #3254 (closed) and #3267 seems to be similar as they also happend with recent kernels and the settings are similar (4k + high refresh rate). However the dmesg output looks quite different.
Brief summary of the problem:
I'm having troubles with >=linux-6.7.6. Since linux-6.7.6, gpu output doesn't work anymore when starting the system. (not even a single line from the kernel booting). However, the system boots normally and i can connect via ssh. Furthermore, when connected via ssh, radeontop also reports 100% memory utilization and i can hear the fan going up to fullspeed.
Now this happens since linux-6.7.6. linux-6.7.5 is still fine and it's also what i'm using at the moment. I've also tested linux-6.7.7 and yesterday linux-6.8.0. All of them have the same problem. There must be some changes between 6.7.5 and 6.7.6 which causes these errors.
Some more information: I'm using Sway-1.9 as my DM (Wayland) with mesa-24.0.3
dmesg output from 6.8.0 will be attached.
Hardware description:
- CPU: AMD Ryzen 9 7950X3D 16-Core Processor
- GPU:
03:00.0 VGA compatible controller [0300]: Advanced Micro Devices, Inc. [AMD/ATI] Navi 31 [Radeon RX 7900 XT/7900 XTX] [1002:744c] (rev c8)
15:00.0 VGA compatible controller [0300]: Advanced Micro Devices, Inc. [AMD/ATI] Raphael [1002:164e] (rev c9)
- System Memory: 64GB
- Display(s):
# ddcutil detect
Display 1
I2C bus: /dev/i2c-6
DRM connector: card0-DP-1
EDID synopsis:
Mfg id: ACR - Acer Technologies
Model: X32 FP
Product code: 2521 (0x09d9)
Serial number: 1246006353G00
Binary serial number: 610272821 (0x24600635)
Manufacture year: 2022, Week: 46
VCP version: 2.2
Display 2
I2C bus: /dev/i2c-7
DRM connector: card0-DP-2
EDID synopsis:
Mfg id: ACR - Acer Technologies
Model: XB283K KV
Product code: 2381 (0x094d)
Serial number: 120918FA54200
Binary serial number: 546410405 (0x20918fa5)
Manufacture year: 2022, Week: 9
VCP version: 2.2
# swaymsg -t get_outputs:
(removed available modes)
Output DP-2 'Acer Technologies XB283K KV 120918FA54200'
Current mode: 3840x2160 @ 144.004 Hz
Power: on
Position: 3840,-1150
Scale factor: 1.000000
Scale filter: nearest
Subpixel hinting: unknown
Transform: 90
Workspace: 2
Max render time: off
Adaptive sync: disabled
Output DP-1 'Acer Technologies X32 FP 1246006353G00' (focused)
Current mode: 3840x2160 @ 143.999 Hz
Power: on
Position: 0,0
Scale factor: 1.000000
Scale filter: nearest
Subpixel hinting: unknown
Transform: normal
Workspace: 1
Max render time: off
Adaptive sync: disabled
- Type of Display Connection: both are connected via DisplayPort
System information:
- Distro name and Version: Gentoo 2.14 ~amd64
- Kernel version: Linux x2 6.7.5-gentoo #1 (closed) SMP PREEMPT_DYNAMIC Sat Feb 17 00:00:11 CET 2024 x86_64 AMD Ryzen 9 7950X3D 16-Core Processor AuthenticAMD GNU/Linux
- Custom kernel: official gentoo-sources: sys-kernel/gentoo-sources-6.7.5
- AMD official driver version: N/A
How to reproduce the issue:
This is pretty easy to reproduce. I only need to update to a kernel newer then 6.7.5 and reboot the system. After rebooting, no output will appear (not even from the kernel itself, before logging in into sway)
Log files (for system lockups / game freezes / crashes)
- dmesg.log from running linux-6.8.0 dmesg.log
- dmesg.log from running linux-6.7.5 (working kernel) dmesg-6.7.5.log
I've also made an bug report at gentoo: https://bugs.gentoo.org/927085 I haven't made a kernel bisect yet but can do (since it's easy to reproduce). Please let me know if this would be desirable.