Bug 216806 - [Raven Ridge] console disappears after framebuffer device loads
Summary: [Raven Ridge] console disappears after framebuffer device loads
Status: CLOSED ANSWERED
Alias: None
Product: Drivers
Classification: Unclassified
Component: Video(DRI - non Intel) (show other bugs)
Hardware: AMD Linux
: P1 normal
Assignee: drivers_video-dri
URL:
Keywords:
Depends on:
Blocks:
 
Reported: 2022-12-13 20:37 UTC by Balazs Vinarz
Modified: 2023-09-30 14:13 UTC (History)
1 user (show)

See Also:
Kernel Version: 6.0.8
Subsystem:
Regression: No
Bisected commit-id:


Attachments
dmesg - debug drm, no dc (82.06 KB, text/plain)
2022-12-13 20:37 UTC, Balazs Vinarz
Details

Description Balazs Vinarz 2022-12-13 20:37:31 UTC
Created attachment 303402 [details]
dmesg - debug drm, no dc

I have an HP c645 Chromebook with Ryzen 3 3250C APU.
Everytime, when the FB/DRM loads and vgaconsole get's deactivated, the console freezes. I tried many common kernel parameters so far with no luck.
By default it freezes back in the boot sequence,
using amdgpu.dc=0 it freezes after the tty shows the login text.
I attached the logs with debug drm flags, with disabling the Display Core driver, the last message is some BT device load message and right after that I found the DRM load.
Comment 1 Alex Deucher 2022-12-13 20:42:12 UTC
please make sure CONFIG_DRM_FBDEV_EMULATION is set in your kernel config.
Comment 2 Balazs Vinarz 2022-12-14 15:02:11 UTC
Hello Alex,
it's set:
$ zgrep DRM_FBDEV /proc/config.gz 
CONFIG_DRM_FBDEV_EMULATION=y
CONFIG_DRM_FBDEV_OVERALLOC=100
I just got a hint to use the iommu=pt, but that one didn't work too.
Comment 3 Balazs Vinarz 2022-12-14 15:08:52 UTC
It looks like this: https://www.youtube.com/watch?v=8oUUBjmo0_I
Comment 4 Balazs Vinarz 2022-12-21 13:56:56 UTC
Just some more information to share here. I gave it another try, and the console  freezes exactly at the moment when drm enables the kernel modesetting. 
[    6.329371] [drm] amdgpu kernel modesetting enabled.
With nomodesetting it works like it should, the vgaconsole remain active and useable.
Do we know somebody who has some knowledge to understand the DRM logs?
Otherwise these lines are pretty strange for me:
[    6.330490] amdgpu: Ignoring ACPI CRAT on non-APU system
.
.
.
[    6.515889] [drm] failed to load ucode RLC_RESTORE_LIST_CNTL(0x28) 
[    6.515892] [drm] psp gfx command LOAD_IP_FW(0x6) failed and response status is (0xFFFF300F)
[    6.516529] [drm] failed to load ucode RLC_RESTORE_LIST_GPM_MEM(0x29) 
[    6.516531] [drm] psp gfx command LOAD_IP_FW(0x6) failed and response status is (0xFFFF000F)
[    6.517064] [drm] failed to load ucode RLC_RESTORE_LIST_SRM_MEM(0x2A) 
[    6.517066] [drm] psp gfx command LOAD_IP_FW(0x6) failed and response status is (0xFFFF000F)
.
.
.
[    6.549311] [drm] psp gfx command LOAD_ASD(0x4) failed and response status is (0x34)
[    6.549317] amdgpu 0000:04:00.0: amdgpu: RAS: optional ras ta ucode is not available
[    6.550582] [drm] psp gfx command LOAD_TA(0x1) failed and response status is (0x34)
[    6.551810] [drm] psp gfx command LOAD_TA(0x1) failed and response status is (0x34)
Comment 5 Balazs Vinarz 2023-09-29 16:44:46 UTC
The latest 6.5.5-arch1-1 kernel just turned the display into a full white screen with high brightness.
Comment 6 Artem S. Tashkinov 2023-09-30 10:33:25 UTC
Please repost here:

https://gitlab.freedesktop.org/drm/amd/-/issues
Comment 7 Balazs Vinarz 2023-09-30 14:13:20 UTC
It turned out with the latest kernel the amd_iommu=off did the trick.
https://gitlab.freedesktop.org/drm/amd/-/issues/2884

Note You need to log in before you can comment on or make changes to this bug.