AMDGPU failing to load with Radeon 550.

lastgosumatt · 2021-12-21 09:03:57

Hello everyone!

I am using ThinkPad e480 with dual GPU - Intel and Radeon 550.
Till recently everything was fine, now the amdgpu is failing to load.

I tried kernels:
linux
linux-zen
linux-lts

Firmware:
linux-firmware-git

Also, I tried to set amdgpu.dc=0.

lspci -k | grep -EA3 'VGA|3D|Display' output

 
00:02.0 VGA compatible controller: Intel Corporation UHD Graphics 620 (rev 07)
	Subsystem: Lenovo Device 5069
	Kernel driver in use: i915
	Kernel modules: i915
--
02:00.0 Display controller: Advanced Micro Devices, Inc. [AMD/ATI] Lexa PRO [Radeon 540/540X/550/550X / RX 540X/550/550X] (rev c0)
	Subsystem: Lenovo Device 5069
	Kernel driver in use: amdgpu
	Kernel modules: amdgpu

sudo dmesg | grep amdgpu output

[    0.000000] Command line: BOOT_IMAGE=/vmlinuz-linux-zen root=UUID=5f5ff9a0-4483-44c1-a24e-7a82c2ea3a1e rw loglevel=3 quiet video=2560x1440@144 amdgpu.dc=0
[    0.064665] Kernel command line: BOOT_IMAGE=/vmlinuz-linux-zen root=UUID=5f5ff9a0-4483-44c1-a24e-7a82c2ea3a1e rw loglevel=3 quiet video=2560x1440@144 amdgpu.dc=0
[    1.367454] [drm] amdgpu kernel modesetting enabled.
[    1.367631] amdgpu: CRAT table not found
[    1.367633] amdgpu: Virtual CRAT table created for CPU
[    1.367641] amdgpu: Topology: Add CPU node
[    1.367847] amdgpu 0000:02:00.0: amdgpu: Trusted Memory Zone (TMZ) feature not supported
[    1.420121] amdgpu 0000:02:00.0: amdgpu: Fetched VBIOS from ATRM
[    1.420123] amdgpu: ATOM BIOS: BR36111.001
[    1.430895] amdgpu 0000:02:00.0: amdgpu: VRAM: 595593216M 0x000000F400000000 - 0x000238F47FFFFFFF (595593216M used)
[    1.430898] amdgpu 0000:02:00.0: amdgpu: GART: 256M 0x0000000000000000 - 0x000000000FFFFFFF
[    1.430926] [drm] amdgpu: 595593216M of VRAM memory ready
[    1.430927] [drm] amdgpu: 11825M of GTT memory ready.
[    1.433444] amdgpu: hwmgr_sw_init smu backed is polaris10_smu
[    3.799300] amdgpu:
[    6.158804] amdgpu: SMU load firmware failed
[    6.158805] amdgpu: fw load failed
[    6.158806] amdgpu: smu firmware loading failed
[    6.158807] amdgpu 0000:02:00.0: amdgpu: amdgpu_device_ip_init failed
[    6.158809] amdgpu 0000:02:00.0: amdgpu: Fatal error during GPU init
[    6.158811] amdgpu 0000:02:00.0: amdgpu: amdgpu: finishing device.
[    6.160352] amdgpu: probe of 0000:02:00.0 failed with error -22
[    6.161047]  amdgpu_uvd_suspend+0x178/0x240 [amdgpu 128c7f93e579f45a030666be14546912cafb82ac]
[    6.161358]  uvd_v6_0_sw_fini+0x10/0x80 [amdgpu 128c7f93e579f45a030666be14546912cafb82ac]
[    6.161552]  amdgpu_device_fini_sw+0xb6/0x2f0 [amdgpu 128c7f93e579f45a030666be14546912cafb82ac]
[    6.161726]  amdgpu_driver_release_kms+0x12/0x30 [amdgpu 128c7f93e579f45a030666be14546912cafb82ac]
[    6.162342] Modules linked in: amdgpu(+) gpu_sched drm_ttm_helper ttm

Thank you for any suggestion.

V1del · 2021-12-21 09:59:19

When did you last actively update that linux-firmware-git package ? Or did you install that as a reaction to this issue or did you do that earlier? Does changing to the repo provided linux-firmware fix this?

lastgosumatt · 2021-12-21 10:10:34

It was reaction for the problem, so I tried both linux-firmware and linux-firmware-git.

V1del · 2021-12-21 10:14:25

Well a firmware loading error points to an issue in loading the firmwaree, maybe instead of trying a newer version you could test a older version of the linux-firmware package? Compare your pacman.log between a point you know it was still working and when it stopped and check the package/linux/linux-firmware changes happening in that timeframe.

lastgosumatt · 2021-12-21 11:49:31

I tried to downgrade both mesa and linux-firmware till my system stopped loading.

Then I used my old live-usb and got this message, which was not there before (same I started to have recently on linux and linux-zen, on linux-lts is all good (except for amdgpu)), sorry I don't know how to copy live usb log: https://drive.google.com/file/d/12ZdVZr … sp=sharing

And shutdown hangs on:
https://drive.google.com/file/d/14WeKlV … sp=sharing

It is the same for live usb, linux and linux-zen.

So I thought maybe it somehow related to amdgpu issue, and maybe it is hardware problem?

Also, because linux can't detect GPU temperature, the fan is always on max speed. But in bios is all ok.

Last edited by lastgosumatt (2021-12-21 12:35:42)

lastgosumatt · 2021-12-23 04:02:41

I don't know what happened exactly, but now it is working.

DRI_PRIME=1 glxinfo | grep "OpenGL renderer"
OpenGL renderer string: AMD Radeon 500 Series (POLARIS12, DRM 3.40.0, 5.10.88-1-lts, LLVM 13.0.0)

Last edited by lastgosumatt (2021-12-23 04:05:24)

seth · 2021-12-23 09:20:21

Some possible causes:
* https://wiki.archlinux.org/title/Kernel … _KMS_start
* Dual boot? 3rd link below.

Arch Linux

#1 2021-12-21 09:03:57

AMDGPU failing to load with Radeon 550.

#2 2021-12-21 09:59:19

Re: AMDGPU failing to load with Radeon 550.

#3 2021-12-21 10:10:34

Re: AMDGPU failing to load with Radeon 550.

#4 2021-12-21 10:14:25

Re: AMDGPU failing to load with Radeon 550.

#5 2021-12-21 11:49:31

Re: AMDGPU failing to load with Radeon 550.

#6 2021-12-23 04:02:41

Re: AMDGPU failing to load with Radeon 550.

#7 2021-12-23 09:20:21

Re: AMDGPU failing to load with Radeon 550.

Board footer