Bug 211981

Summary:	AMDGPU: without cable attached videocard shuts down and does not wake up again, log prints "[drm] Cannot find any crtc or sizes"
Product:	Drivers	Reporter:	raul (raulvior.bcn)
Component:	Video(DRI - non Intel)	Assignee:	drivers_video-dri
Status:	NEW ---
Severity:	normal	CC:	alexdeucher
Priority:	P1
Hardware:	All
OS:	Linux
Kernel Version:	5.10.10	Subsystem:
Regression:	No	Bisected commit-id:
Attachments:	Dmesg when no output is attached. Framebuffer not initalized.

Description raul 2021-02-27 13:51:51 UTC

Created attachment 295505 [details]
Dmesg when no output is attached. Framebuffer not initalized.

Linux boot shows the message in the title of this bug report if the videocard does not have any output attached. The computer boots up and when Linux starts loading, the card shuts down: fans stop spinning and the LED of the logo turns off. It's a Sapphire Radeon RX 580. If I attach the cable past this point the card does not turn on again. Pressing keys on the keyboard, moving the mouse or turning off and on the monitor does not awake the GPU.

> amdgpu 0000:27:00.0: amdgpu: SE 4, SH per SE 1, CU per SH 9, active_cu_number
> 36
> amdgpu 0000:27:00.0: [drm] Cannot find any crtc or sizes
> [drm] Initialized amdgpu 3.40.0 20150101 for 0000:27:00.0 on minor 0


If there's an output device attached to the videocard, then it prints this text:
> amdgpu 0000:27:00.0: amdgpu: SE 4, SH per SE 1, CU per SH 9, active_cu_number
> 36
> [drm] fb mappable at 0xE0BEB000
> [drm] vram apper at 0xE0000000
> [drm] size 7299072
> [drm] fb depth is 24
> [drm]    pitch is 6912
> fbcon: amdgpudrmfb (fb0) is primary device
> EDAC amd64: F17h detected (node 0).
> EDAC amd64: Node 0: DRAM ECC disabled.
> Console: switching to colour frame buffer device 210x65
> amdgpu 0000:27:00.0: [drm] fb0: amdgpudrmfb frame buffer device
> [drm] Initialized amdgpu 3.40.0 20150101 for 0000:27:00.0 on minor 0

Comment 1 Alex Deucher 2021-03-02 15:43:10 UTC

Does setting amdgpu.runpm=0 fix it?

Comment 2 raul 2021-03-03 14:55:36 UTC

(In reply to Alex Deucher from comment #1)
> Does setting amdgpu.runpm=0 fix it?

Yes, it does.